diff --git "a/checkpoint-2000/trainer_state.json" "b/checkpoint-2000/trainer_state.json" new file mode 100644--- /dev/null +++ "b/checkpoint-2000/trainer_state.json" @@ -0,0 +1,28033 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 0.6373486297004461, + "eval_steps": 500, + "global_step": 2000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "completion_length": 108.203125, + "epoch": 0.00031867431485022306, + "grad_norm": 29.355905532836914, + "kl": 0.0, + "learning_rate": 9.996813256851498e-07, + "loss": 0.0, + "reward": 0.990570604801178, + "reward_std": 0.5569084882736206, + "rewards/answer_reward": 0.140625, + "rewards/format_reward_gqa": 0.71875, + "rewards/iou_glue_reward": 0.13119560480117798, + "step": 1 + }, + { + "completion_length": 130.765625, + "epoch": 0.0006373486297004461, + "grad_norm": 54.51008987426758, + "kl": 0.00274658203125, + "learning_rate": 9.993626513702996e-07, + "loss": 0.0001, + "reward": 0.6362947225570679, + "reward_std": 0.46545886993408203, + "rewards/format_reward_tg": 0.484375, + "rewards/iou_timestamp_reward": 0.15191972255706787, + "rewards/pad": 0.0, + "step": 2 + }, + { + "completion_length": 106.125, + "epoch": 0.0009560229445506692, + "grad_norm": 27.533252716064453, + "kl": 0.006683349609375, + "learning_rate": 9.990439770554494e-07, + "loss": 0.0003, + "reward": 1.0761747360229492, + "reward_std": 0.4438322186470032, + "rewards/pad": 0.046875, + "rewards/tracking_format_reward": 0.796875, + "rewards/tracking_iou_reward": 0.232424795627594, + "step": 3 + }, + { + "completion_length": 18.84375, + "epoch": 0.0012746972594008922, + "grad_norm": 27.014707565307617, + "kl": 0.00341796875, + "learning_rate": 9.98725302740599e-07, + "loss": 0.0001, + "reward": 1.1961885690689087, + "reward_std": 0.2257140427827835, + "rewards/format_reward_tg": 0.96875, + "rewards/iou_timestamp_reward": 0.2274385243654251, + "rewards/pad": 0.0, + "step": 4 + }, + { + "completion_length": 63.953125, + "epoch": 0.0015933715742511153, + "grad_norm": 45.08222579956055, + "kl": 0.0023345947265625, + "learning_rate": 9.984066284257488e-07, + "loss": 0.0001, + "reward": 1.518608570098877, + "reward_std": 0.3346985876560211, + "rewards/answer_reward": 0.328125, + "rewards/format_reward_gqa": 0.90625, + "rewards/iou_glue_reward": 0.2842335104942322, + "step": 5 + }, + { + "completion_length": 80.90625, + "epoch": 0.0019120458891013384, + "grad_norm": 14.024462699890137, + "kl": 0.0037841796875, + "learning_rate": 9.980879541108986e-07, + "loss": 0.0002, + "reward": 1.2835232019424438, + "reward_std": 0.21698495745658875, + "rewards/pad": 0.140625, + "rewards/tracking_format_reward": 0.9375, + "rewards/tracking_iou_reward": 0.20539820194244385, + "step": 6 + }, + { + "completion_length": 74.609375, + "epoch": 0.0022307202039515616, + "grad_norm": 24.76462745666504, + "kl": 0.004638671875, + "learning_rate": 9.977692797960484e-07, + "loss": 0.0002, + "reward": 1.3342989683151245, + "reward_std": 0.33252787590026855, + "rewards/format_reward_tg": 0.953125, + "rewards/iou_timestamp_reward": 0.17804892361164093, + "rewards/pad": 0.203125, + "step": 7 + }, + { + "completion_length": 94.6875, + "epoch": 0.0025493945188017845, + "grad_norm": 111.181884765625, + "kl": 0.00567626953125, + "learning_rate": 9.974506054811982e-07, + "loss": 0.0002, + "reward": 1.2611994743347168, + "reward_std": 0.356869637966156, + "rewards/format_reward_tg": 0.890625, + "rewards/iou_timestamp_reward": 0.37057459354400635, + "rewards/pad": 0.0, + "step": 8 + }, + { + "completion_length": 92.203125, + "epoch": 0.0028680688336520078, + "grad_norm": 30.41423988342285, + "kl": 0.017578125, + "learning_rate": 9.97131931166348e-07, + "loss": 0.0007, + "reward": 1.3457612991333008, + "reward_std": 0.3058345317840576, + "rewards/format_reward_tg": 0.875, + "rewards/iou_timestamp_reward": 0.36138632893562317, + "rewards/pad": 0.109375, + "step": 9 + }, + { + "completion_length": 104.0625, + "epoch": 0.0031867431485022306, + "grad_norm": 44.20106887817383, + "kl": 0.006317138671875, + "learning_rate": 9.968132568514978e-07, + "loss": 0.0003, + "reward": 1.3545455932617188, + "reward_std": 0.3300294280052185, + "rewards/answer_reward": 0.1875, + "rewards/format_reward_gqa": 0.921875, + "rewards/iou_glue_reward": 0.2451706826686859, + "step": 10 + }, + { + "completion_length": 120.953125, + "epoch": 0.003505417463352454, + "grad_norm": 63.420772552490234, + "kl": 0.01275634765625, + "learning_rate": 9.964945825366476e-07, + "loss": 0.0005, + "reward": 1.15547776222229, + "reward_std": 0.2632618248462677, + "rewards/format_reward_tg": 0.90625, + "rewards/iou_timestamp_reward": 0.24922770261764526, + "rewards/pad": 0.0, + "step": 11 + }, + { + "completion_length": 124.828125, + "epoch": 0.0038240917782026767, + "grad_norm": 7.296823978424072, + "kl": 0.004913330078125, + "learning_rate": 9.961759082217972e-07, + "loss": 0.0002, + "reward": 1.311105728149414, + "reward_std": 0.36636123061180115, + "rewards/pad": 0.03125, + "rewards/tracking_format_reward": 0.953125, + "rewards/tracking_iou_reward": 0.32673075795173645, + "step": 12 + }, + { + "completion_length": 75.421875, + "epoch": 0.0041427660930529, + "grad_norm": 32.08348083496094, + "kl": 0.017578125, + "learning_rate": 9.95857233906947e-07, + "loss": 0.0007, + "reward": 1.307035207748413, + "reward_std": 0.2346232533454895, + "rewards/format_reward_tg": 0.9375, + "rewards/iou_timestamp_reward": 0.27578526735305786, + "rewards/pad": 0.09375, + "step": 13 + }, + { + "completion_length": 120.078125, + "epoch": 0.004461440407903123, + "grad_norm": 17.275251388549805, + "kl": 0.01446533203125, + "learning_rate": 9.955385595920968e-07, + "loss": 0.0006, + "reward": 1.2805306911468506, + "reward_std": 0.2489503175020218, + "rewards/format_reward_tg": 0.953125, + "rewards/iou_timestamp_reward": 0.32740581035614014, + "rewards/pad": 0.0, + "step": 14 + }, + { + "completion_length": 99.359375, + "epoch": 0.004780114722753346, + "grad_norm": 17.721607208251953, + "kl": 0.009521484375, + "learning_rate": 9.952198852772466e-07, + "loss": 0.0004, + "reward": 1.4107003211975098, + "reward_std": 0.3282482922077179, + "rewards/answer_reward": 0.109375, + "rewards/format_reward_gqa": 0.96875, + "rewards/iou_glue_reward": 0.33257538080215454, + "step": 15 + }, + { + "completion_length": 41.578125, + "epoch": 0.005098789037603569, + "grad_norm": 17.887115478515625, + "kl": 0.0361328125, + "learning_rate": 9.949012109623964e-07, + "loss": 0.0014, + "reward": 1.4330815076828003, + "reward_std": 0.26210516691207886, + "rewards/format_reward_tg": 0.96875, + "rewards/iou_timestamp_reward": 0.3393315076828003, + "rewards/pad": 0.125, + "step": 16 + }, + { + "completion_length": 93.46875, + "epoch": 0.005417463352453792, + "grad_norm": 49.0717658996582, + "kl": 0.019287109375, + "learning_rate": 9.945825366475462e-07, + "loss": 0.0008, + "reward": 1.46000337600708, + "reward_std": 0.2627638578414917, + "rewards/format_reward_tg": 0.953125, + "rewards/iou_timestamp_reward": 0.2725033462047577, + "rewards/pad": 0.234375, + "step": 17 + }, + { + "completion_length": 128.984375, + "epoch": 0.0057361376673040155, + "grad_norm": 16.0430965423584, + "kl": 0.018798828125, + "learning_rate": 9.94263862332696e-07, + "loss": 0.0008, + "reward": 1.2777396440505981, + "reward_std": 0.3498499095439911, + "rewards/pad": 0.109375, + "rewards/tracking_format_reward": 0.90625, + "rewards/tracking_iou_reward": 0.262114554643631, + "step": 18 + }, + { + "completion_length": 120.3125, + "epoch": 0.006054811982154238, + "grad_norm": 21.588294982910156, + "kl": 0.0830078125, + "learning_rate": 9.939451880178459e-07, + "loss": 0.0033, + "reward": 1.2359671592712402, + "reward_std": 0.22133731842041016, + "rewards/format_reward_tg": 0.96875, + "rewards/iou_timestamp_reward": 0.2672172784805298, + "rewards/pad": 0.0, + "step": 19 + }, + { + "completion_length": 117.171875, + "epoch": 0.006373486297004461, + "grad_norm": 9.763358116149902, + "kl": 0.0220947265625, + "learning_rate": 9.936265137029955e-07, + "loss": 0.0009, + "reward": 1.3035460710525513, + "reward_std": 0.17414405941963196, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.3191710114479065, + "rewards/pad": 0.0, + "step": 20 + }, + { + "completion_length": 143.859375, + "epoch": 0.006692160611854685, + "grad_norm": 22.552978515625, + "kl": 0.030517578125, + "learning_rate": 9.933078393881453e-07, + "loss": 0.0012, + "reward": 1.4276816844940186, + "reward_std": 0.2420690506696701, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.96875, + "rewards/tracking_iou_reward": 0.45893165469169617, + "step": 21 + }, + { + "completion_length": 116.59375, + "epoch": 0.007010834926704908, + "grad_norm": 19.098342895507812, + "kl": 0.04345703125, + "learning_rate": 9.92989165073295e-07, + "loss": 0.0017, + "reward": 1.4236531257629395, + "reward_std": 0.24127241969108582, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.31427815556526184, + "step": 22 + }, + { + "completion_length": 124.640625, + "epoch": 0.007329509241555131, + "grad_norm": 13.978249549865723, + "kl": 0.0301513671875, + "learning_rate": 9.926704907584449e-07, + "loss": 0.0012, + "reward": 1.263721227645874, + "reward_std": 0.2416178286075592, + "rewards/pad": 0.015625, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.2637212872505188, + "step": 23 + }, + { + "completion_length": 95.109375, + "epoch": 0.0076481835564053535, + "grad_norm": 18.538372039794922, + "kl": 0.034912109375, + "learning_rate": 9.923518164435945e-07, + "loss": 0.0014, + "reward": 1.498464822769165, + "reward_std": 0.18835774064064026, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.2640899121761322, + "rewards/pad": 0.234375, + "step": 24 + }, + { + "completion_length": 139.75, + "epoch": 0.007966857871255577, + "grad_norm": 15.361570358276367, + "kl": 0.017822265625, + "learning_rate": 9.920331421287443e-07, + "loss": 0.0007, + "reward": 1.37433922290802, + "reward_std": 0.13456077873706818, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.37433922290802, + "step": 25 + }, + { + "completion_length": 71.703125, + "epoch": 0.0082855321861058, + "grad_norm": 192.22190856933594, + "kl": 0.0279541015625, + "learning_rate": 9.91714467813894e-07, + "loss": 0.0011, + "reward": 1.4138457775115967, + "reward_std": 0.19462621212005615, + "rewards/pad": 0.140625, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.2732207179069519, + "step": 26 + }, + { + "completion_length": 114.0, + "epoch": 0.008604206500956023, + "grad_norm": 39.867950439453125, + "kl": 0.05029296875, + "learning_rate": 9.91395793499044e-07, + "loss": 0.002, + "reward": 1.3192813396453857, + "reward_std": 0.20698723196983337, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.33490636944770813, + "step": 27 + }, + { + "completion_length": 69.296875, + "epoch": 0.008922880815806247, + "grad_norm": 25.988426208496094, + "kl": 0.06640625, + "learning_rate": 9.910771191841937e-07, + "loss": 0.0026, + "reward": 1.4162707328796387, + "reward_std": 0.16801214218139648, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.41627073287963867, + "step": 28 + }, + { + "completion_length": 95.875, + "epoch": 0.009241555130656469, + "grad_norm": 11.73469352722168, + "kl": 0.04638671875, + "learning_rate": 9.907584448693435e-07, + "loss": 0.0019, + "reward": 1.5614410638809204, + "reward_std": 0.2746657133102417, + "rewards/answer_reward": 0.234375, + "rewards/format_reward_gqa": 0.984375, + "rewards/iou_glue_reward": 0.3426910638809204, + "step": 29 + }, + { + "completion_length": 92.140625, + "epoch": 0.009560229445506692, + "grad_norm": 27.682270050048828, + "kl": 0.06494140625, + "learning_rate": 9.904397705544933e-07, + "loss": 0.0026, + "reward": 1.2468478679656982, + "reward_std": 0.19129014015197754, + "rewards/format_reward_tg": 0.96875, + "rewards/iou_timestamp_reward": 0.27809786796569824, + "rewards/pad": 0.0, + "step": 30 + }, + { + "completion_length": 94.25, + "epoch": 0.009878903760356916, + "grad_norm": 35.90079116821289, + "kl": 0.043212890625, + "learning_rate": 9.90121096239643e-07, + "loss": 0.0017, + "reward": 1.382009744644165, + "reward_std": 0.19762729108333588, + "rewards/pad": 0.078125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3038847744464874, + "step": 31 + }, + { + "completion_length": 118.140625, + "epoch": 0.010197578075207138, + "grad_norm": 15.5087251663208, + "kl": 0.0191650390625, + "learning_rate": 9.898024219247927e-07, + "loss": 0.0008, + "reward": 1.4223272800445557, + "reward_std": 0.18307654559612274, + "rewards/answer_reward": 0.140625, + "rewards/format_reward_gqa": 0.984375, + "rewards/iou_glue_reward": 0.2973273694515228, + "step": 32 + }, + { + "completion_length": 94.015625, + "epoch": 0.010516252390057362, + "grad_norm": 18.41666030883789, + "kl": 0.0654296875, + "learning_rate": 9.894837476099425e-07, + "loss": 0.0026, + "reward": 1.390166163444519, + "reward_std": 0.11873508989810944, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.39016619324684143, + "step": 33 + }, + { + "completion_length": 144.484375, + "epoch": 0.010834926704907584, + "grad_norm": 23.03544807434082, + "kl": 0.049560546875, + "learning_rate": 9.891650732950923e-07, + "loss": 0.002, + "reward": 1.2460135221481323, + "reward_std": 0.13103874027729034, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.26163846254348755, + "step": 34 + }, + { + "completion_length": 43.703125, + "epoch": 0.011153601019757807, + "grad_norm": 74.67973327636719, + "kl": 0.2021484375, + "learning_rate": 9.888463989802421e-07, + "loss": 0.0081, + "reward": 1.3883543014526367, + "reward_std": 0.2638249099254608, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.4039793610572815, + "rewards/pad": 0.0, + "step": 35 + }, + { + "completion_length": 72.484375, + "epoch": 0.011472275334608031, + "grad_norm": 18.90091323852539, + "kl": 0.034912109375, + "learning_rate": 9.88527724665392e-07, + "loss": 0.0014, + "reward": 1.5052143335342407, + "reward_std": 0.19993306696414948, + "rewards/answer_reward": 0.140625, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.36458930373191833, + "step": 36 + }, + { + "completion_length": 93.5, + "epoch": 0.011790949649458253, + "grad_norm": 42.74220657348633, + "kl": 0.04931640625, + "learning_rate": 9.882090503505418e-07, + "loss": 0.002, + "reward": 1.364863634109497, + "reward_std": 0.1607465147972107, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.27111363410949707, + "rewards/pad": 0.09375, + "step": 37 + }, + { + "completion_length": 121.765625, + "epoch": 0.012109623964308477, + "grad_norm": 12.138788223266602, + "kl": 0.0322265625, + "learning_rate": 9.878903760356916e-07, + "loss": 0.0013, + "reward": 1.5847129821777344, + "reward_std": 0.12323808670043945, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.3347129821777344, + "step": 38 + }, + { + "completion_length": 45.453125, + "epoch": 0.0124282982791587, + "grad_norm": 94.58944702148438, + "kl": 0.126953125, + "learning_rate": 9.875717017208412e-07, + "loss": 0.0051, + "reward": 1.4467828273773193, + "reward_std": 0.22572225332260132, + "rewards/answer_reward": 0.046875, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.39990779757499695, + "step": 39 + }, + { + "completion_length": 93.390625, + "epoch": 0.012746972594008922, + "grad_norm": 26.145246505737305, + "kl": 0.053466796875, + "learning_rate": 9.87253027405991e-07, + "loss": 0.0021, + "reward": 1.4706435203552246, + "reward_std": 0.1406538188457489, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.470643550157547, + "step": 40 + }, + { + "completion_length": 45.984375, + "epoch": 0.013065646908859146, + "grad_norm": 76.75983428955078, + "kl": 0.0732421875, + "learning_rate": 9.869343530911408e-07, + "loss": 0.0029, + "reward": 1.4271475076675415, + "reward_std": 0.16546779870986938, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3021474778652191, + "rewards/pad": 0.125, + "step": 41 + }, + { + "completion_length": 73.953125, + "epoch": 0.01338432122370937, + "grad_norm": 21.900846481323242, + "kl": 0.058349609375, + "learning_rate": 9.866156787762906e-07, + "loss": 0.0023, + "reward": 1.5581902265548706, + "reward_std": 0.10484491288661957, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.30819016695022583, + "step": 42 + }, + { + "completion_length": 71.515625, + "epoch": 0.013702995538559592, + "grad_norm": 36.57977294921875, + "kl": 0.06689453125, + "learning_rate": 9.862970044614404e-07, + "loss": 0.0027, + "reward": 1.5820021629333496, + "reward_std": 0.21441596746444702, + "rewards/answer_reward": 0.21875, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.36325210332870483, + "step": 43 + }, + { + "completion_length": 146.640625, + "epoch": 0.014021669853409816, + "grad_norm": 18.229549407958984, + "kl": 0.0289306640625, + "learning_rate": 9.859783301465902e-07, + "loss": 0.0012, + "reward": 1.2833337783813477, + "reward_std": 0.10058476030826569, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.28333374857902527, + "rewards/pad": 0.0, + "step": 44 + }, + { + "completion_length": 70.28125, + "epoch": 0.014340344168260038, + "grad_norm": 15.984444618225098, + "kl": 0.06640625, + "learning_rate": 9.8565965583174e-07, + "loss": 0.0026, + "reward": 1.5217663049697876, + "reward_std": 0.21203118562698364, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.537391185760498, + "rewards/pad": 0.0, + "step": 45 + }, + { + "completion_length": 70.953125, + "epoch": 0.014659018483110261, + "grad_norm": 24.116294860839844, + "kl": 0.054443359375, + "learning_rate": 9.853409815168898e-07, + "loss": 0.0022, + "reward": 1.4242897033691406, + "reward_std": 0.1250244379043579, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3149147033691406, + "rewards/pad": 0.109375, + "step": 46 + }, + { + "completion_length": 96.171875, + "epoch": 0.014977692797960485, + "grad_norm": 10.824599266052246, + "kl": 0.04931640625, + "learning_rate": 9.850223072020394e-07, + "loss": 0.002, + "reward": 1.3301568031311035, + "reward_std": 0.09323126077651978, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3301568627357483, + "rewards/pad": 0.0, + "step": 47 + }, + { + "completion_length": 69.328125, + "epoch": 0.015296367112810707, + "grad_norm": 54.09326171875, + "kl": 0.0859375, + "learning_rate": 9.847036328871892e-07, + "loss": 0.0034, + "reward": 1.5022802352905273, + "reward_std": 0.11983383446931839, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5022802352905273, + "rewards/pad": 0.0, + "step": 48 + }, + { + "completion_length": 94.390625, + "epoch": 0.01561504142766093, + "grad_norm": 20.41078758239746, + "kl": 0.04345703125, + "learning_rate": 9.84384958572339e-07, + "loss": 0.0017, + "reward": 1.5264580249786377, + "reward_std": 0.11588997393846512, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5264579653739929, + "step": 49 + }, + { + "completion_length": 70.96875, + "epoch": 0.015933715742511154, + "grad_norm": 11.669675827026367, + "kl": 0.052001953125, + "learning_rate": 9.840662842574888e-07, + "loss": 0.0021, + "reward": 1.5469558238983154, + "reward_std": 0.1315864473581314, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5469557642936707, + "rewards/pad": 0.0, + "step": 50 + }, + { + "completion_length": 44.578125, + "epoch": 0.016252390057361378, + "grad_norm": 49.088687896728516, + "kl": 0.05322265625, + "learning_rate": 9.837476099426386e-07, + "loss": 0.0021, + "reward": 1.5957996845245361, + "reward_std": 0.13811209797859192, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.47079968452453613, + "rewards/pad": 0.125, + "step": 51 + }, + { + "completion_length": 146.546875, + "epoch": 0.0165710643722116, + "grad_norm": 7.562353610992432, + "kl": 0.02490234375, + "learning_rate": 9.834289356277884e-07, + "loss": 0.001, + "reward": 1.329128623008728, + "reward_std": 0.17861336469650269, + "rewards/format_reward_tg": 0.96875, + "rewards/iou_timestamp_reward": 0.360378623008728, + "rewards/pad": 0.0, + "step": 52 + }, + { + "completion_length": 120.109375, + "epoch": 0.016889738687061822, + "grad_norm": 20.31233024597168, + "kl": 0.048583984375, + "learning_rate": 9.831102613129382e-07, + "loss": 0.0019, + "reward": 1.5360966920852661, + "reward_std": 0.195622980594635, + "rewards/pad": 0.0625, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4735966622829437, + "step": 53 + }, + { + "completion_length": 95.796875, + "epoch": 0.017208413001912046, + "grad_norm": 18.229900360107422, + "kl": 0.048583984375, + "learning_rate": 9.82791586998088e-07, + "loss": 0.0019, + "reward": 1.4456595182418823, + "reward_std": 0.1510319858789444, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.38315948843955994, + "rewards/pad": 0.0625, + "step": 54 + }, + { + "completion_length": 71.78125, + "epoch": 0.01752708731676227, + "grad_norm": 41.68214416503906, + "kl": 0.1240234375, + "learning_rate": 9.824729126832376e-07, + "loss": 0.0049, + "reward": 1.5753334760665894, + "reward_std": 0.12371576577425003, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4503334164619446, + "rewards/pad": 0.125, + "step": 55 + }, + { + "completion_length": 45.34375, + "epoch": 0.017845761631612493, + "grad_norm": 19.551204681396484, + "kl": 0.0927734375, + "learning_rate": 9.821542383683875e-07, + "loss": 0.0037, + "reward": 1.3601016998291016, + "reward_std": 0.09858176857233047, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.36010172963142395, + "rewards/pad": 0.0, + "step": 56 + }, + { + "completion_length": 132.375, + "epoch": 0.018164435946462717, + "grad_norm": 25.195838928222656, + "kl": 0.039794921875, + "learning_rate": 9.818355640535373e-07, + "loss": 0.0016, + "reward": 1.487829566001892, + "reward_std": 0.11461742967367172, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.3784545361995697, + "step": 57 + }, + { + "completion_length": 97.546875, + "epoch": 0.018483110261312937, + "grad_norm": 38.37260818481445, + "kl": 0.0478515625, + "learning_rate": 9.81516889738687e-07, + "loss": 0.0019, + "reward": 1.5829166173934937, + "reward_std": 0.1692945510149002, + "rewards/pad": 0.15625, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4266665577888489, + "step": 58 + }, + { + "completion_length": 97.65625, + "epoch": 0.01880178457616316, + "grad_norm": 28.56045150756836, + "kl": 0.0625, + "learning_rate": 9.811982154238367e-07, + "loss": 0.0025, + "reward": 1.3472189903259277, + "reward_std": 0.22873595356941223, + "rewards/pad": 0.046875, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.31596893072128296, + "step": 59 + }, + { + "completion_length": 150.375, + "epoch": 0.019120458891013385, + "grad_norm": 9.740062713623047, + "kl": 0.05224609375, + "learning_rate": 9.808795411089865e-07, + "loss": 0.0021, + "reward": 1.4982421398162842, + "reward_std": 0.14206960797309875, + "rewards/pad": 0.109375, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.40449216961860657, + "step": 60 + }, + { + "completion_length": 149.0, + "epoch": 0.019439133205863608, + "grad_norm": 13.651617050170898, + "kl": 0.037841796875, + "learning_rate": 9.805608667941363e-07, + "loss": 0.0015, + "reward": 1.5308611392974854, + "reward_std": 0.1005527451634407, + "rewards/pad": 0.09375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.43711116909980774, + "step": 61 + }, + { + "completion_length": 96.15625, + "epoch": 0.019757807520713832, + "grad_norm": 17.998046875, + "kl": 0.0810546875, + "learning_rate": 9.80242192479286e-07, + "loss": 0.0032, + "reward": 1.4157941341400146, + "reward_std": 0.08607807010412216, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.41579413414001465, + "rewards/pad": 0.0, + "step": 62 + }, + { + "completion_length": 71.890625, + "epoch": 0.020076481835564052, + "grad_norm": 137.67347717285156, + "kl": 0.1318359375, + "learning_rate": 9.799235181644359e-07, + "loss": 0.0053, + "reward": 1.411577582359314, + "reward_std": 0.1643928587436676, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3803275525569916, + "rewards/pad": 0.03125, + "step": 63 + }, + { + "completion_length": 95.40625, + "epoch": 0.020395156150414276, + "grad_norm": 12.385643005371094, + "kl": 0.059814453125, + "learning_rate": 9.796048438495857e-07, + "loss": 0.0024, + "reward": 1.5682718753814697, + "reward_std": 0.17435622215270996, + "rewards/pad": 0.078125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4901469349861145, + "step": 64 + }, + { + "completion_length": 97.96875, + "epoch": 0.0207138304652645, + "grad_norm": 29.116153717041016, + "kl": 0.10791015625, + "learning_rate": 9.792861695347355e-07, + "loss": 0.0043, + "reward": 1.6022427082061768, + "reward_std": 0.10109306126832962, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.47724267840385437, + "rewards/pad": 0.125, + "step": 65 + }, + { + "completion_length": 96.171875, + "epoch": 0.021032504780114723, + "grad_norm": 24.117246627807617, + "kl": 0.09423828125, + "learning_rate": 9.78967495219885e-07, + "loss": 0.0038, + "reward": 1.3648128509521484, + "reward_std": 0.09639433771371841, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3648129105567932, + "rewards/pad": 0.0, + "step": 66 + }, + { + "completion_length": 69.65625, + "epoch": 0.021351179094964947, + "grad_norm": 89.97595977783203, + "kl": 0.09228515625, + "learning_rate": 9.78648820905035e-07, + "loss": 0.0037, + "reward": 1.284343957901001, + "reward_std": 0.14161378145217896, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.2999690771102905, + "step": 67 + }, + { + "completion_length": 74.125, + "epoch": 0.021669853409815167, + "grad_norm": 28.639371871948242, + "kl": 0.0859375, + "learning_rate": 9.783301465901847e-07, + "loss": 0.0034, + "reward": 1.4897210597991943, + "reward_std": 0.1362878680229187, + "rewards/pad": 0.015625, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.47409600019454956, + "step": 68 + }, + { + "completion_length": 97.53125, + "epoch": 0.02198852772466539, + "grad_norm": 39.00709533691406, + "kl": 0.09814453125, + "learning_rate": 9.780114722753345e-07, + "loss": 0.0039, + "reward": 1.5572547912597656, + "reward_std": 0.0959487184882164, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5572547912597656, + "rewards/pad": 0.0, + "step": 69 + }, + { + "completion_length": 147.46875, + "epoch": 0.022307202039515615, + "grad_norm": 24.02517318725586, + "kl": 0.0537109375, + "learning_rate": 9.776927979604843e-07, + "loss": 0.0022, + "reward": 1.6853044033050537, + "reward_std": 0.06498823314905167, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4353044033050537, + "step": 70 + }, + { + "completion_length": 124.421875, + "epoch": 0.02262587635436584, + "grad_norm": 12.759849548339844, + "kl": 0.0771484375, + "learning_rate": 9.773741236456341e-07, + "loss": 0.0031, + "reward": 1.4843802452087402, + "reward_std": 0.07213138788938522, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4843802750110626, + "rewards/pad": 0.0, + "step": 71 + }, + { + "completion_length": 125.28125, + "epoch": 0.022944550669216062, + "grad_norm": 33.39959716796875, + "kl": 0.0859375, + "learning_rate": 9.77055449330784e-07, + "loss": 0.0035, + "reward": 1.5397720336914062, + "reward_std": 0.07022380828857422, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4147719740867615, + "step": 72 + }, + { + "completion_length": 121.890625, + "epoch": 0.023263224984066286, + "grad_norm": 17.097454071044922, + "kl": 0.0537109375, + "learning_rate": 9.767367750159337e-07, + "loss": 0.0022, + "reward": 1.3611290454864502, + "reward_std": 0.07813698053359985, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3611290752887726, + "step": 73 + }, + { + "completion_length": 73.34375, + "epoch": 0.023581899298916506, + "grad_norm": 32.879371643066406, + "kl": 0.07177734375, + "learning_rate": 9.764181007010833e-07, + "loss": 0.0029, + "reward": 1.70439875125885, + "reward_std": 0.1615179181098938, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4543987810611725, + "step": 74 + }, + { + "completion_length": 178.828125, + "epoch": 0.02390057361376673, + "grad_norm": 42.670570373535156, + "kl": 0.031982421875, + "learning_rate": 9.760994263862331e-07, + "loss": 0.0013, + "reward": 1.333021879196167, + "reward_std": 0.05501440539956093, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.33302196860313416, + "step": 75 + }, + { + "completion_length": 147.0625, + "epoch": 0.024219247928616953, + "grad_norm": 7.011380195617676, + "kl": 0.057861328125, + "learning_rate": 9.75780752071383e-07, + "loss": 0.0023, + "reward": 1.564239740371704, + "reward_std": 0.06192938610911369, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4392397403717041, + "step": 76 + }, + { + "completion_length": 98.046875, + "epoch": 0.024537922243467177, + "grad_norm": 18.060924530029297, + "kl": 0.0810546875, + "learning_rate": 9.754620777565328e-07, + "loss": 0.0032, + "reward": 1.668859839439392, + "reward_std": 0.1618225872516632, + "rewards/pad": 0.234375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4344848394393921, + "step": 77 + }, + { + "completion_length": 72.84375, + "epoch": 0.0248565965583174, + "grad_norm": 108.19387817382812, + "kl": 0.0908203125, + "learning_rate": 9.751434034416826e-07, + "loss": 0.0036, + "reward": 1.6406117677688599, + "reward_std": 0.10990151762962341, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.40623676776885986, + "rewards/pad": 0.234375, + "step": 78 + }, + { + "completion_length": 18.921875, + "epoch": 0.02517527087316762, + "grad_norm": 44.06772232055664, + "kl": 0.1240234375, + "learning_rate": 9.748247291268324e-07, + "loss": 0.005, + "reward": 1.5604881048202515, + "reward_std": 0.13976165652275085, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.43548810482025146, + "rewards/pad": 0.125, + "step": 79 + }, + { + "completion_length": 46.390625, + "epoch": 0.025493945188017845, + "grad_norm": 43.962501525878906, + "kl": 0.09326171875, + "learning_rate": 9.745060548119822e-07, + "loss": 0.0037, + "reward": 1.5611708164215088, + "reward_std": 0.10798977315425873, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.43617090582847595, + "rewards/pad": 0.125, + "step": 80 + }, + { + "completion_length": 18.734375, + "epoch": 0.02581261950286807, + "grad_norm": 21.721975326538086, + "kl": 0.08203125, + "learning_rate": 9.74187380497132e-07, + "loss": 0.0033, + "reward": 1.5381839275360107, + "reward_std": 0.10050734877586365, + "rewards/answer_reward": 0.0, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5381839871406555, + "step": 81 + }, + { + "completion_length": 70.0, + "epoch": 0.026131293817718292, + "grad_norm": 57.080387115478516, + "kl": 0.09912109375, + "learning_rate": 9.738687061822816e-07, + "loss": 0.004, + "reward": 1.4988303184509277, + "reward_std": 0.10620397329330444, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4988303780555725, + "rewards/pad": 0.0, + "step": 82 + }, + { + "completion_length": 125.953125, + "epoch": 0.026449968132568516, + "grad_norm": 89.73050689697266, + "kl": 0.0615234375, + "learning_rate": 9.735500318674314e-07, + "loss": 0.0025, + "reward": 1.4511033296585083, + "reward_std": 0.06983766704797745, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4511033296585083, + "rewards/pad": 0.0, + "step": 83 + }, + { + "completion_length": 68.9375, + "epoch": 0.02676864244741874, + "grad_norm": 52.88443374633789, + "kl": 0.1171875, + "learning_rate": 9.732313575525812e-07, + "loss": 0.0047, + "reward": 1.5206553936004639, + "reward_std": 0.07981610298156738, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3956553339958191, + "step": 84 + }, + { + "completion_length": 124.8125, + "epoch": 0.02708731676226896, + "grad_norm": 30.62926483154297, + "kl": 0.06982421875, + "learning_rate": 9.72912683237731e-07, + "loss": 0.0028, + "reward": 1.3461110591888428, + "reward_std": 0.07935292273759842, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.34611111879348755, + "rewards/pad": 0.0, + "step": 85 + }, + { + "completion_length": 97.4375, + "epoch": 0.027405991077119184, + "grad_norm": 21.218219757080078, + "kl": 0.09326171875, + "learning_rate": 9.725940089228808e-07, + "loss": 0.0037, + "reward": 1.3955249786376953, + "reward_std": 0.06659087538719177, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.39552485942840576, + "rewards/pad": 0.0, + "step": 86 + }, + { + "completion_length": 45.296875, + "epoch": 0.027724665391969407, + "grad_norm": 34.96338653564453, + "kl": 0.14453125, + "learning_rate": 9.722753346080306e-07, + "loss": 0.0058, + "reward": 1.6848323345184326, + "reward_std": 0.09776704013347626, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.6848323941230774, + "step": 87 + }, + { + "completion_length": 43.28125, + "epoch": 0.02804333970681963, + "grad_norm": 39.88020324707031, + "kl": 0.1259765625, + "learning_rate": 9.719566602931804e-07, + "loss": 0.005, + "reward": 1.414764404296875, + "reward_std": 0.12384898960590363, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4147644639015198, + "rewards/pad": 0.0, + "step": 88 + }, + { + "completion_length": 72.03125, + "epoch": 0.028362014021669855, + "grad_norm": 75.94140625, + "kl": 0.1044921875, + "learning_rate": 9.716379859783302e-07, + "loss": 0.0042, + "reward": 1.4549014568328857, + "reward_std": 0.10878540575504303, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4549015164375305, + "step": 89 + }, + { + "completion_length": 45.140625, + "epoch": 0.028680688336520075, + "grad_norm": 31.220703125, + "kl": 0.1328125, + "learning_rate": 9.713193116634798e-07, + "loss": 0.0053, + "reward": 1.4554842710494995, + "reward_std": 0.12406035512685776, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4554842710494995, + "step": 90 + }, + { + "completion_length": 96.9375, + "epoch": 0.0289993626513703, + "grad_norm": 17.68728256225586, + "kl": 0.1064453125, + "learning_rate": 9.710006373486296e-07, + "loss": 0.0043, + "reward": 1.5634052753448486, + "reward_std": 0.12161899358034134, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5634053349494934, + "rewards/pad": 0.0, + "step": 91 + }, + { + "completion_length": 73.71875, + "epoch": 0.029318036966220522, + "grad_norm": 32.400978088378906, + "kl": 0.138671875, + "learning_rate": 9.706819630337794e-07, + "loss": 0.0056, + "reward": 1.432898759841919, + "reward_std": 0.17899563908576965, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.35477375984191895, + "rewards/pad": 0.078125, + "step": 92 + }, + { + "completion_length": 125.6875, + "epoch": 0.029636711281070746, + "grad_norm": 47.505828857421875, + "kl": 0.08740234375, + "learning_rate": 9.703632887189293e-07, + "loss": 0.0035, + "reward": 1.3663368225097656, + "reward_std": 0.02895917370915413, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3663369119167328, + "rewards/pad": 0.0, + "step": 93 + }, + { + "completion_length": 72.234375, + "epoch": 0.02995538559592097, + "grad_norm": 66.74494934082031, + "kl": 0.10791015625, + "learning_rate": 9.70044614404079e-07, + "loss": 0.0043, + "reward": 1.5021812915802002, + "reward_std": 0.048981934785842896, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.37718141078948975, + "step": 94 + }, + { + "completion_length": 70.796875, + "epoch": 0.030274059910771194, + "grad_norm": 99.29576110839844, + "kl": 0.0859375, + "learning_rate": 9.697259400892289e-07, + "loss": 0.0034, + "reward": 1.5479052066802979, + "reward_std": 0.14039847254753113, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5479051470756531, + "rewards/pad": 0.0, + "step": 95 + }, + { + "completion_length": 18.96875, + "epoch": 0.030592734225621414, + "grad_norm": 39.55213165283203, + "kl": 0.1103515625, + "learning_rate": 9.694072657743787e-07, + "loss": 0.0044, + "reward": 1.5478798151016235, + "reward_std": 0.14118030667304993, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.43850481510162354, + "rewards/pad": 0.109375, + "step": 96 + }, + { + "completion_length": 98.5625, + "epoch": 0.030911408540471638, + "grad_norm": 14.295442581176758, + "kl": 0.0986328125, + "learning_rate": 9.690885914595283e-07, + "loss": 0.004, + "reward": 1.566028118133545, + "reward_std": 0.06352254748344421, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.44102805852890015, + "step": 97 + }, + { + "completion_length": 43.828125, + "epoch": 0.03123008285532186, + "grad_norm": 18.267284393310547, + "kl": 0.1201171875, + "learning_rate": 9.68769917144678e-07, + "loss": 0.0048, + "reward": 1.4908885955810547, + "reward_std": 0.0876583606004715, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.49088847637176514, + "rewards/pad": 0.0, + "step": 98 + }, + { + "completion_length": 125.5625, + "epoch": 0.03154875717017208, + "grad_norm": 92.24208068847656, + "kl": 0.0576171875, + "learning_rate": 9.684512428298279e-07, + "loss": 0.0023, + "reward": 1.5934380292892456, + "reward_std": 0.05329665541648865, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4684380292892456, + "rewards/pad": 0.125, + "step": 99 + }, + { + "completion_length": 45.984375, + "epoch": 0.03186743148502231, + "grad_norm": 182.80059814453125, + "kl": 0.1005859375, + "learning_rate": 9.681325685149777e-07, + "loss": 0.004, + "reward": 1.5487216711044312, + "reward_std": 0.16690734028816223, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5018466711044312, + "rewards/pad": 0.046875, + "step": 100 + }, + { + "completion_length": 96.59375, + "epoch": 0.03218610579987253, + "grad_norm": 47.996280670166016, + "kl": 0.08544921875, + "learning_rate": 9.678138942001273e-07, + "loss": 0.0034, + "reward": 1.4153481721878052, + "reward_std": 0.11429637670516968, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.415348082780838, + "rewards/pad": 0.0, + "step": 101 + }, + { + "completion_length": 73.4375, + "epoch": 0.032504780114722756, + "grad_norm": 129.4193878173828, + "kl": 0.09765625, + "learning_rate": 9.67495219885277e-07, + "loss": 0.0039, + "reward": 1.408902883529663, + "reward_std": 0.10345868021249771, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.2995278239250183, + "rewards/pad": 0.109375, + "step": 102 + }, + { + "completion_length": 98.625, + "epoch": 0.032823454429572976, + "grad_norm": 26.774145126342773, + "kl": 0.072265625, + "learning_rate": 9.67176545570427e-07, + "loss": 0.0029, + "reward": 1.6206049919128418, + "reward_std": 0.0844387486577034, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4956050515174866, + "step": 103 + }, + { + "completion_length": 96.5625, + "epoch": 0.0331421287444232, + "grad_norm": 105.86109161376953, + "kl": 0.08984375, + "learning_rate": 9.668578712555767e-07, + "loss": 0.0036, + "reward": 1.3795945644378662, + "reward_std": 0.06641143560409546, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.25459450483322144, + "step": 104 + }, + { + "completion_length": 99.796875, + "epoch": 0.033460803059273424, + "grad_norm": 13.890254020690918, + "kl": 0.08984375, + "learning_rate": 9.665391969407265e-07, + "loss": 0.0036, + "reward": 1.606649398803711, + "reward_std": 0.13339820504188538, + "rewards/answer_reward": 0.15625, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.45039939880371094, + "step": 105 + }, + { + "completion_length": 147.921875, + "epoch": 0.033779477374123644, + "grad_norm": 7.908030033111572, + "kl": 0.05419921875, + "learning_rate": 9.662205226258763e-07, + "loss": 0.0022, + "reward": 1.4531539678573608, + "reward_std": 0.027786817401647568, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4531539976596832, + "step": 106 + }, + { + "completion_length": 72.859375, + "epoch": 0.03409815168897387, + "grad_norm": 32.176780700683594, + "kl": 0.173828125, + "learning_rate": 9.659018483110261e-07, + "loss": 0.0069, + "reward": 1.6003332138061523, + "reward_std": 0.11983513832092285, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.47533318400382996, + "rewards/pad": 0.125, + "step": 107 + }, + { + "completion_length": 151.421875, + "epoch": 0.03441682600382409, + "grad_norm": 50.097869873046875, + "kl": 0.054443359375, + "learning_rate": 9.65583173996176e-07, + "loss": 0.0022, + "reward": 1.3763339519500732, + "reward_std": 0.08023016154766083, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3763338625431061, + "rewards/pad": 0.0, + "step": 108 + }, + { + "completion_length": 74.28125, + "epoch": 0.03473550031867431, + "grad_norm": 54.777732849121094, + "kl": 0.06689453125, + "learning_rate": 9.652644996813255e-07, + "loss": 0.0027, + "reward": 1.739059329032898, + "reward_std": 0.11556351184844971, + "rewards/answer_reward": 0.265625, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.47343435883522034, + "step": 109 + }, + { + "completion_length": 45.03125, + "epoch": 0.03505417463352454, + "grad_norm": 51.98408508300781, + "kl": 0.1953125, + "learning_rate": 9.649458253664753e-07, + "loss": 0.0078, + "reward": 1.497527837753296, + "reward_std": 0.11776410043239594, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4975278973579407, + "rewards/pad": 0.0, + "step": 110 + }, + { + "completion_length": 97.546875, + "epoch": 0.03537284894837476, + "grad_norm": 193.61141967773438, + "kl": 0.10791015625, + "learning_rate": 9.646271510516251e-07, + "loss": 0.0043, + "reward": 1.5598293542861938, + "reward_std": 0.07324188947677612, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5598292946815491, + "step": 111 + }, + { + "completion_length": 47.328125, + "epoch": 0.035691523263224986, + "grad_norm": 27.857454299926758, + "kl": 0.10595703125, + "learning_rate": 9.64308476736775e-07, + "loss": 0.0042, + "reward": 1.9212465286254883, + "reward_std": 0.12968027591705322, + "rewards/answer_reward": 0.375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5462466478347778, + "step": 112 + }, + { + "completion_length": 124.796875, + "epoch": 0.036010197578075206, + "grad_norm": 18.940677642822266, + "kl": 0.0732421875, + "learning_rate": 9.639898024219248e-07, + "loss": 0.0029, + "reward": 1.4641155004501343, + "reward_std": 0.09638328850269318, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3391154706478119, + "step": 113 + }, + { + "completion_length": 97.8125, + "epoch": 0.036328871892925434, + "grad_norm": 31.145307540893555, + "kl": 0.0908203125, + "learning_rate": 9.636711281070746e-07, + "loss": 0.0036, + "reward": 1.6159067153930664, + "reward_std": 0.07679077982902527, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6159066557884216, + "rewards/pad": 0.0, + "step": 114 + }, + { + "completion_length": 100.859375, + "epoch": 0.036647546207775654, + "grad_norm": 26.970640182495117, + "kl": 0.0849609375, + "learning_rate": 9.633524537922244e-07, + "loss": 0.0034, + "reward": 1.561607003211975, + "reward_std": 0.1283750832080841, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.43660691380500793, + "step": 115 + }, + { + "completion_length": 73.5, + "epoch": 0.036966220522625874, + "grad_norm": 19.1359920501709, + "kl": 0.0927734375, + "learning_rate": 9.630337794773742e-07, + "loss": 0.0037, + "reward": 1.2844882011413574, + "reward_std": 0.15992632508277893, + "rewards/pad": 0.0625, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.22198832035064697, + "step": 116 + }, + { + "completion_length": 118.59375, + "epoch": 0.0372848948374761, + "grad_norm": 30.564661026000977, + "kl": 0.0966796875, + "learning_rate": 9.627151051625238e-07, + "loss": 0.0039, + "reward": 1.3892366886138916, + "reward_std": 0.09500478953123093, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.389236718416214, + "rewards/pad": 0.0, + "step": 117 + }, + { + "completion_length": 97.96875, + "epoch": 0.03760356915232632, + "grad_norm": 38.80213928222656, + "kl": 0.1904296875, + "learning_rate": 9.623964308476736e-07, + "loss": 0.0076, + "reward": 1.5466018915176392, + "reward_std": 0.06538740545511246, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4216019809246063, + "step": 118 + }, + { + "completion_length": 100.90625, + "epoch": 0.03792224346717655, + "grad_norm": 219.59561157226562, + "kl": 0.08837890625, + "learning_rate": 9.620777565328234e-07, + "loss": 0.0035, + "reward": 1.541412591934204, + "reward_std": 0.08481577038764954, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4164126217365265, + "step": 119 + }, + { + "completion_length": 121.953125, + "epoch": 0.03824091778202677, + "grad_norm": 21.892406463623047, + "kl": 0.1416015625, + "learning_rate": 9.617590822179732e-07, + "loss": 0.0057, + "reward": 1.3950331211090088, + "reward_std": 0.04945141077041626, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.39503324031829834, + "step": 120 + }, + { + "completion_length": 70.25, + "epoch": 0.03855959209687699, + "grad_norm": 25.756412506103516, + "kl": 0.08154296875, + "learning_rate": 9.61440407903123e-07, + "loss": 0.0033, + "reward": 1.7107679843902588, + "reward_std": 0.13024044036865234, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4607679545879364, + "rewards/pad": 0.25, + "step": 121 + }, + { + "completion_length": 98.953125, + "epoch": 0.038878266411727216, + "grad_norm": 58.248538970947266, + "kl": 0.119140625, + "learning_rate": 9.611217335882728e-07, + "loss": 0.0047, + "reward": 1.3422318696975708, + "reward_std": 0.08788175880908966, + "rewards/pad": 0.015625, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3266068398952484, + "step": 122 + }, + { + "completion_length": 125.6875, + "epoch": 0.03919694072657744, + "grad_norm": 21.87851905822754, + "kl": 0.08056640625, + "learning_rate": 9.608030592734226e-07, + "loss": 0.0032, + "reward": 1.7041547298431396, + "reward_std": 0.10210521519184113, + "rewards/answer_reward": 0.359375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.3447796404361725, + "step": 123 + }, + { + "completion_length": 73.796875, + "epoch": 0.039515615041427664, + "grad_norm": 15.746590614318848, + "kl": 0.09765625, + "learning_rate": 9.604843849585724e-07, + "loss": 0.0039, + "reward": 1.4994181394577026, + "reward_std": 0.0809524804353714, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.24941813945770264, + "step": 124 + }, + { + "completion_length": 44.90625, + "epoch": 0.039834289356277884, + "grad_norm": 103.32615661621094, + "kl": 0.1845703125, + "learning_rate": 9.60165710643722e-07, + "loss": 0.0074, + "reward": 1.421347737312317, + "reward_std": 0.07847784459590912, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4213476777076721, + "rewards/pad": 0.0, + "step": 125 + }, + { + "completion_length": 18.03125, + "epoch": 0.040152963671128104, + "grad_norm": 58.665218353271484, + "kl": 0.1748046875, + "learning_rate": 9.598470363288718e-07, + "loss": 0.007, + "reward": 1.4241862297058105, + "reward_std": 0.15667256712913513, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.42418625950813293, + "rewards/pad": 0.0, + "step": 126 + }, + { + "completion_length": 74.640625, + "epoch": 0.04047163798597833, + "grad_norm": 21.96946907043457, + "kl": 0.09716796875, + "learning_rate": 9.595283620140216e-07, + "loss": 0.0039, + "reward": 1.6786143779754639, + "reward_std": 0.14680366218090057, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3348643481731415, + "rewards/pad": 0.34375, + "step": 127 + }, + { + "completion_length": 124.390625, + "epoch": 0.04079031230082855, + "grad_norm": 92.82318115234375, + "kl": 0.09912109375, + "learning_rate": 9.592096876991714e-07, + "loss": 0.004, + "reward": 1.4549717903137207, + "reward_std": 0.05895674601197243, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4549717903137207, + "rewards/pad": 0.0, + "step": 128 + }, + { + "completion_length": 98.109375, + "epoch": 0.04110898661567878, + "grad_norm": 46.42023849487305, + "kl": 0.150390625, + "learning_rate": 9.588910133843212e-07, + "loss": 0.006, + "reward": 1.5591964721679688, + "reward_std": 0.056670114398002625, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5591964721679688, + "rewards/pad": 0.0, + "step": 129 + }, + { + "completion_length": 123.140625, + "epoch": 0.041427660930529, + "grad_norm": 23.843793869018555, + "kl": 0.0849609375, + "learning_rate": 9.58572339069471e-07, + "loss": 0.0034, + "reward": 1.438868761062622, + "reward_std": 0.07457967847585678, + "rewards/answer_reward": 0.0, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.43886876106262207, + "step": 130 + }, + { + "completion_length": 124.53125, + "epoch": 0.04174633524537922, + "grad_norm": 34.21272659301758, + "kl": 0.06494140625, + "learning_rate": 9.582536647546209e-07, + "loss": 0.0026, + "reward": 1.582020878791809, + "reward_std": 0.09043914824724197, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4570208489894867, + "rewards/pad": 0.125, + "step": 131 + }, + { + "completion_length": 72.96875, + "epoch": 0.04206500956022945, + "grad_norm": 24.28638458251953, + "kl": 0.10546875, + "learning_rate": 9.579349904397705e-07, + "loss": 0.0042, + "reward": 1.610738754272461, + "reward_std": 0.10264147818088531, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4857388436794281, + "rewards/pad": 0.125, + "step": 132 + }, + { + "completion_length": 98.0625, + "epoch": 0.04238368387507967, + "grad_norm": 23.711889266967773, + "kl": 0.091796875, + "learning_rate": 9.576163161249203e-07, + "loss": 0.0037, + "reward": 1.6860997676849365, + "reward_std": 0.06541076302528381, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5610998272895813, + "step": 133 + }, + { + "completion_length": 126.109375, + "epoch": 0.042702358189929894, + "grad_norm": 18.623653411865234, + "kl": 0.06689453125, + "learning_rate": 9.5729764181007e-07, + "loss": 0.0027, + "reward": 1.5307762622833252, + "reward_std": 0.030945513397455215, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4057762920856476, + "step": 134 + }, + { + "completion_length": 125.59375, + "epoch": 0.043021032504780114, + "grad_norm": 17.338851928710938, + "kl": 0.09765625, + "learning_rate": 9.569789674952199e-07, + "loss": 0.0039, + "reward": 1.5449299812316895, + "reward_std": 0.032470062375068665, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5449299812316895, + "step": 135 + }, + { + "completion_length": 44.5, + "epoch": 0.043339706819630335, + "grad_norm": 58.93782043457031, + "kl": 0.208984375, + "learning_rate": 9.566602931803697e-07, + "loss": 0.0084, + "reward": 1.482414722442627, + "reward_std": 0.11973986029624939, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.49803969264030457, + "step": 136 + }, + { + "completion_length": 124.65625, + "epoch": 0.04365838113448056, + "grad_norm": 68.23857879638672, + "kl": 0.1171875, + "learning_rate": 9.563416188655195e-07, + "loss": 0.0047, + "reward": 1.4399811029434204, + "reward_std": 0.03828233852982521, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.31498122215270996, + "step": 137 + }, + { + "completion_length": 45.65625, + "epoch": 0.04397705544933078, + "grad_norm": 124.44922637939453, + "kl": 0.14453125, + "learning_rate": 9.56022944550669e-07, + "loss": 0.0058, + "reward": 1.6428006887435913, + "reward_std": 0.07791727781295776, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5178006887435913, + "rewards/pad": 0.125, + "step": 138 + }, + { + "completion_length": 98.3125, + "epoch": 0.04429572976418101, + "grad_norm": 15.683511734008789, + "kl": 0.09619140625, + "learning_rate": 9.557042702358189e-07, + "loss": 0.0038, + "reward": 1.4232029914855957, + "reward_std": 0.06578314304351807, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.2982029318809509, + "step": 139 + }, + { + "completion_length": 101.09375, + "epoch": 0.04461440407903123, + "grad_norm": 46.468170166015625, + "kl": 0.0947265625, + "learning_rate": 9.553855959209687e-07, + "loss": 0.0038, + "reward": 1.7962223291397095, + "reward_std": 0.06880475580692291, + "rewards/pad": 0.375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.42122238874435425, + "step": 140 + }, + { + "completion_length": 99.25, + "epoch": 0.044933078393881457, + "grad_norm": 24.893571853637695, + "kl": 0.08447265625, + "learning_rate": 9.550669216061185e-07, + "loss": 0.0034, + "reward": 1.7566449642181396, + "reward_std": 0.07449847459793091, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5066449642181396, + "step": 141 + }, + { + "completion_length": 126.484375, + "epoch": 0.04525175270873168, + "grad_norm": 47.23504638671875, + "kl": 0.05615234375, + "learning_rate": 9.547482472912683e-07, + "loss": 0.0022, + "reward": 1.7445024251937866, + "reward_std": 0.08240990340709686, + "rewards/pad": 0.328125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4163774847984314, + "step": 142 + }, + { + "completion_length": 151.34375, + "epoch": 0.0455704270235819, + "grad_norm": 24.97036361694336, + "kl": 0.07275390625, + "learning_rate": 9.544295729764181e-07, + "loss": 0.0029, + "reward": 1.490628957748413, + "reward_std": 0.03826358914375305, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4906289279460907, + "step": 143 + }, + { + "completion_length": 150.546875, + "epoch": 0.045889101338432124, + "grad_norm": 1052.7979736328125, + "kl": 0.099609375, + "learning_rate": 9.541108986615677e-07, + "loss": 0.004, + "reward": 1.446044921875, + "reward_std": 0.054064881056547165, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4460448920726776, + "rewards/pad": 0.0, + "step": 144 + }, + { + "completion_length": 123.046875, + "epoch": 0.046207775653282344, + "grad_norm": 26.211153030395508, + "kl": 0.1005859375, + "learning_rate": 9.537922243467175e-07, + "loss": 0.004, + "reward": 1.42512047290802, + "reward_std": 0.034479159861803055, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4251205325126648, + "step": 145 + }, + { + "completion_length": 71.53125, + "epoch": 0.04652644996813257, + "grad_norm": 69.134033203125, + "kl": 0.1298828125, + "learning_rate": 9.534735500318673e-07, + "loss": 0.0052, + "reward": 1.553114652633667, + "reward_std": 0.12183529883623123, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.45936471223831177, + "rewards/pad": 0.09375, + "step": 146 + }, + { + "completion_length": 97.984375, + "epoch": 0.04684512428298279, + "grad_norm": 56.753135681152344, + "kl": 0.115234375, + "learning_rate": 9.531548757170171e-07, + "loss": 0.0046, + "reward": 1.6361443996429443, + "reward_std": 0.056340161710977554, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5111443996429443, + "step": 147 + }, + { + "completion_length": 46.15625, + "epoch": 0.04716379859783301, + "grad_norm": 90.41157531738281, + "kl": 1.4453125, + "learning_rate": 9.528362014021669e-07, + "loss": 0.0577, + "reward": 1.532861590385437, + "reward_std": 0.14235638082027435, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 0.984375, + "rewards/iou_glue_reward": 0.4234865605831146, + "step": 148 + }, + { + "completion_length": 45.6875, + "epoch": 0.04748247291268324, + "grad_norm": 108.94004821777344, + "kl": 0.11865234375, + "learning_rate": 9.525175270873167e-07, + "loss": 0.0048, + "reward": 1.6035866737365723, + "reward_std": 0.09949560463428497, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4785866141319275, + "step": 149 + }, + { + "completion_length": 71.78125, + "epoch": 0.04780114722753346, + "grad_norm": 50.10005187988281, + "kl": 0.1318359375, + "learning_rate": 9.521988527724664e-07, + "loss": 0.0053, + "reward": 1.529174566268921, + "reward_std": 0.09229524433612823, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5291745662689209, + "step": 150 + }, + { + "completion_length": 71.578125, + "epoch": 0.04811982154238369, + "grad_norm": 42.62626266479492, + "kl": 0.11572265625, + "learning_rate": 9.518801784576163e-07, + "loss": 0.0046, + "reward": 1.4902132749557495, + "reward_std": 0.03644484281539917, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4902133345603943, + "rewards/pad": 0.0, + "step": 151 + }, + { + "completion_length": 72.5, + "epoch": 0.04843849585723391, + "grad_norm": 37.46217346191406, + "kl": 0.1318359375, + "learning_rate": 9.515615041427661e-07, + "loss": 0.0053, + "reward": 1.7063305377960205, + "reward_std": 0.08680470287799835, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5813304781913757, + "rewards/pad": 0.125, + "step": 152 + }, + { + "completion_length": 98.09375, + "epoch": 0.04875717017208413, + "grad_norm": 27.439014434814453, + "kl": 0.0849609375, + "learning_rate": 9.512428298279159e-07, + "loss": 0.0034, + "reward": 1.634913444519043, + "reward_std": 0.08088289201259613, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.509913444519043, + "step": 153 + }, + { + "completion_length": 175.203125, + "epoch": 0.049075844486934354, + "grad_norm": 22.50582504272461, + "kl": 0.046630859375, + "learning_rate": 9.509241555130656e-07, + "loss": 0.0019, + "reward": 1.4140303134918213, + "reward_std": 0.0470597930252552, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.41403043270111084, + "step": 154 + }, + { + "completion_length": 72.328125, + "epoch": 0.049394518801784575, + "grad_norm": 165.1951446533203, + "kl": 0.1865234375, + "learning_rate": 9.506054811982154e-07, + "loss": 0.0075, + "reward": 1.6961311101913452, + "reward_std": 0.06627112627029419, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5711310505867004, + "rewards/pad": 0.125, + "step": 155 + }, + { + "completion_length": 149.5, + "epoch": 0.0497131931166348, + "grad_norm": 165.7172393798828, + "kl": 0.1474609375, + "learning_rate": 9.502868068833652e-07, + "loss": 0.0059, + "reward": 1.3272875547409058, + "reward_std": 0.04778099060058594, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.32728755474090576, + "step": 156 + }, + { + "completion_length": 71.9375, + "epoch": 0.05003186743148502, + "grad_norm": 56.9672737121582, + "kl": 0.1044921875, + "learning_rate": 9.499681325685149e-07, + "loss": 0.0042, + "reward": 1.6385527849197388, + "reward_std": 0.055156029760837555, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5135527849197388, + "rewards/pad": 0.125, + "step": 157 + }, + { + "completion_length": 70.9375, + "epoch": 0.05035054174633524, + "grad_norm": 18.57289695739746, + "kl": 0.12158203125, + "learning_rate": 9.496494582536647e-07, + "loss": 0.0049, + "reward": 1.642876148223877, + "reward_std": 0.07265563309192657, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6428762078285217, + "rewards/pad": 0.0, + "step": 158 + }, + { + "completion_length": 49.21875, + "epoch": 0.05066921606118547, + "grad_norm": 31.87525749206543, + "kl": 0.1357421875, + "learning_rate": 9.493307839388145e-07, + "loss": 0.0054, + "reward": 1.5862581729888916, + "reward_std": 0.11294609308242798, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.586258053779602, + "rewards/pad": 0.0, + "step": 159 + }, + { + "completion_length": 120.8125, + "epoch": 0.05098789037603569, + "grad_norm": 28.032197952270508, + "kl": 0.0830078125, + "learning_rate": 9.490121096239643e-07, + "loss": 0.0033, + "reward": 1.4823583364486694, + "reward_std": 0.056141920387744904, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.48235827684402466, + "rewards/pad": 0.0, + "step": 160 + }, + { + "completion_length": 68.859375, + "epoch": 0.05130656469088592, + "grad_norm": 130.83055114746094, + "kl": 0.0751953125, + "learning_rate": 9.48693435309114e-07, + "loss": 0.003, + "reward": 1.520409345626831, + "reward_std": 0.10488361120223999, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.520409345626831, + "step": 161 + }, + { + "completion_length": 68.625, + "epoch": 0.05162523900573614, + "grad_norm": 38.07346725463867, + "kl": 0.1494140625, + "learning_rate": 9.483747609942638e-07, + "loss": 0.006, + "reward": 1.66463303565979, + "reward_std": 0.13315197825431824, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 0.984375, + "rewards/iou_glue_reward": 0.5552579760551453, + "step": 162 + }, + { + "completion_length": 72.765625, + "epoch": 0.05194391332058636, + "grad_norm": 17.296497344970703, + "kl": 0.1904296875, + "learning_rate": 9.480560866794136e-07, + "loss": 0.0076, + "reward": 1.6184468269348145, + "reward_std": 0.07699354737997055, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.36844679713249207, + "step": 163 + }, + { + "completion_length": 98.1875, + "epoch": 0.052262587635436585, + "grad_norm": 30.80031394958496, + "kl": 0.09814453125, + "learning_rate": 9.477374123645634e-07, + "loss": 0.0039, + "reward": 1.640599012374878, + "reward_std": 0.12581594288349152, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5312240123748779, + "rewards/pad": 0.109375, + "step": 164 + }, + { + "completion_length": 148.9375, + "epoch": 0.052581261950286805, + "grad_norm": 63.38698959350586, + "kl": 0.0615234375, + "learning_rate": 9.474187380497131e-07, + "loss": 0.0025, + "reward": 1.4976701736450195, + "reward_std": 0.08813410997390747, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.41954517364501953, + "rewards/pad": 0.078125, + "step": 165 + }, + { + "completion_length": 95.671875, + "epoch": 0.05289993626513703, + "grad_norm": 24.988746643066406, + "kl": 0.09228515625, + "learning_rate": 9.471000637348629e-07, + "loss": 0.0037, + "reward": 1.6525627374649048, + "reward_std": 0.06571772694587708, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.52756267786026, + "step": 166 + }, + { + "completion_length": 205.3125, + "epoch": 0.05321861057998725, + "grad_norm": 7.888558864593506, + "kl": 0.0306396484375, + "learning_rate": 9.467813894200127e-07, + "loss": 0.0012, + "reward": 1.3351364135742188, + "reward_std": 0.07257430255413055, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.35076141357421875, + "step": 167 + }, + { + "completion_length": 72.03125, + "epoch": 0.05353728489483748, + "grad_norm": 531.3428955078125, + "kl": 0.10791015625, + "learning_rate": 9.464627151051625e-07, + "loss": 0.0043, + "reward": 1.7654680013656616, + "reward_std": 0.13501885533332825, + "rewards/pad": 0.234375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5310930013656616, + "step": 168 + }, + { + "completion_length": 18.78125, + "epoch": 0.0538559592096877, + "grad_norm": 35.41008758544922, + "kl": 0.1513671875, + "learning_rate": 9.461440407903123e-07, + "loss": 0.0061, + "reward": 1.6094670295715332, + "reward_std": 0.08094550669193268, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4844670295715332, + "rewards/pad": 0.125, + "step": 169 + }, + { + "completion_length": 68.234375, + "epoch": 0.05417463352453792, + "grad_norm": 80.35694122314453, + "kl": 0.10693359375, + "learning_rate": 9.458253664754621e-07, + "loss": 0.0043, + "reward": 1.7217336893081665, + "reward_std": 0.15385891497135162, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.6123586297035217, + "rewards/pad": 0.125, + "step": 170 + }, + { + "completion_length": 71.3125, + "epoch": 0.05449330783938815, + "grad_norm": 43.178199768066406, + "kl": 0.10400390625, + "learning_rate": 9.455066921606119e-07, + "loss": 0.0042, + "reward": 1.353318214416504, + "reward_std": 0.03510623052716255, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.35331809520721436, + "step": 171 + }, + { + "completion_length": 96.90625, + "epoch": 0.05481198215423837, + "grad_norm": 71.26564025878906, + "kl": 0.08642578125, + "learning_rate": 9.451880178457617e-07, + "loss": 0.0034, + "reward": 1.468902826309204, + "reward_std": 0.06541343033313751, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.46890294551849365, + "rewards/pad": 0.0, + "step": 172 + }, + { + "completion_length": 150.1875, + "epoch": 0.055130656469088594, + "grad_norm": 77.41221618652344, + "kl": 0.06884765625, + "learning_rate": 9.448693435309114e-07, + "loss": 0.0028, + "reward": 1.3299999237060547, + "reward_std": 0.07897648215293884, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.34562498331069946, + "rewards/pad": 0.0, + "step": 173 + }, + { + "completion_length": 45.5, + "epoch": 0.055449330783938815, + "grad_norm": 30.494190216064453, + "kl": 0.119140625, + "learning_rate": 9.445506692160612e-07, + "loss": 0.0048, + "reward": 1.527397871017456, + "reward_std": 0.12671810388565063, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.41802287101745605, + "rewards/pad": 0.125, + "step": 174 + }, + { + "completion_length": 19.90625, + "epoch": 0.055768005098789035, + "grad_norm": 29.110815048217773, + "kl": 0.1376953125, + "learning_rate": 9.44231994901211e-07, + "loss": 0.0055, + "reward": 1.4674365520477295, + "reward_std": 0.08331962674856186, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4674365222454071, + "rewards/pad": 0.0, + "step": 175 + }, + { + "completion_length": 123.265625, + "epoch": 0.05608667941363926, + "grad_norm": 35.95558547973633, + "kl": 0.0791015625, + "learning_rate": 9.439133205863608e-07, + "loss": 0.0032, + "reward": 1.4634273052215576, + "reward_std": 0.06282143294811249, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.46342724561691284, + "step": 176 + }, + { + "completion_length": 97.890625, + "epoch": 0.05640535372848948, + "grad_norm": 26.46843719482422, + "kl": 0.0810546875, + "learning_rate": 9.435946462715104e-07, + "loss": 0.0032, + "reward": 1.5546252727508545, + "reward_std": 0.09996479004621506, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4452502727508545, + "rewards/pad": 0.109375, + "step": 177 + }, + { + "completion_length": 176.015625, + "epoch": 0.05672402804333971, + "grad_norm": 23.104013442993164, + "kl": 0.07373046875, + "learning_rate": 9.432759719566602e-07, + "loss": 0.0029, + "reward": 1.4256776571273804, + "reward_std": 0.038105182349681854, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.42567768692970276, + "step": 178 + }, + { + "completion_length": 72.109375, + "epoch": 0.05704270235818993, + "grad_norm": 152.2291717529297, + "kl": 0.103515625, + "learning_rate": 9.4295729764181e-07, + "loss": 0.0041, + "reward": 1.5971485376358032, + "reward_std": 0.111331045627594, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.425273597240448, + "rewards/pad": 0.171875, + "step": 179 + }, + { + "completion_length": 123.390625, + "epoch": 0.05736137667304015, + "grad_norm": 62.74386978149414, + "kl": 0.146484375, + "learning_rate": 9.426386233269598e-07, + "loss": 0.0059, + "reward": 1.5698823928833008, + "reward_std": 0.08544135093688965, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5698824524879456, + "step": 180 + }, + { + "completion_length": 97.421875, + "epoch": 0.05768005098789038, + "grad_norm": 21.25946807861328, + "kl": 0.0966796875, + "learning_rate": 9.423199490121095e-07, + "loss": 0.0039, + "reward": 1.7297738790512085, + "reward_std": 0.04847279191017151, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.47977393865585327, + "rewards/pad": 0.25, + "step": 181 + }, + { + "completion_length": 19.953125, + "epoch": 0.0579987253027406, + "grad_norm": 72.88169860839844, + "kl": 0.1396484375, + "learning_rate": 9.420012746972593e-07, + "loss": 0.0056, + "reward": 1.551754117012024, + "reward_std": 0.09995436668395996, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3017541766166687, + "rewards/pad": 0.25, + "step": 182 + }, + { + "completion_length": 98.6875, + "epoch": 0.058317399617590825, + "grad_norm": 82.24500274658203, + "kl": 0.07470703125, + "learning_rate": 9.416826003824091e-07, + "loss": 0.003, + "reward": 1.7466256618499756, + "reward_std": 0.0803009569644928, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.49662575125694275, + "rewards/pad": 0.25, + "step": 183 + }, + { + "completion_length": 123.078125, + "epoch": 0.058636073932441045, + "grad_norm": 58.43217849731445, + "kl": 0.09033203125, + "learning_rate": 9.413639260675588e-07, + "loss": 0.0036, + "reward": 1.549285888671875, + "reward_std": 0.05893456190824509, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.549285888671875, + "step": 184 + }, + { + "completion_length": 97.734375, + "epoch": 0.058954748247291265, + "grad_norm": 34.839447021484375, + "kl": 0.1396484375, + "learning_rate": 9.410452517527086e-07, + "loss": 0.0056, + "reward": 1.7135741710662842, + "reward_std": 0.08446040004491806, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5885741710662842, + "step": 185 + }, + { + "completion_length": 71.25, + "epoch": 0.05927342256214149, + "grad_norm": 27.027359008789062, + "kl": 0.125, + "learning_rate": 9.407265774378584e-07, + "loss": 0.005, + "reward": 1.5394039154052734, + "reward_std": 0.10799270868301392, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4300287961959839, + "rewards/pad": 0.109375, + "step": 186 + }, + { + "completion_length": 123.328125, + "epoch": 0.05959209687699171, + "grad_norm": 115.98084259033203, + "kl": 0.12451171875, + "learning_rate": 9.404079031230082e-07, + "loss": 0.005, + "reward": 1.323099136352539, + "reward_std": 0.05522124096751213, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.32309919595718384, + "rewards/pad": 0.0, + "step": 187 + }, + { + "completion_length": 97.640625, + "epoch": 0.05991077119184194, + "grad_norm": 51.958438873291016, + "kl": 0.09619140625, + "learning_rate": 9.40089228808158e-07, + "loss": 0.0039, + "reward": 1.541533350944519, + "reward_std": 0.11370637267827988, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.557158350944519, + "rewards/pad": 0.0, + "step": 188 + }, + { + "completion_length": 148.828125, + "epoch": 0.06022944550669216, + "grad_norm": 33.664268493652344, + "kl": 0.04931640625, + "learning_rate": 9.397705544933078e-07, + "loss": 0.002, + "reward": 1.4180933237075806, + "reward_std": 0.05330347269773483, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.41809332370758057, + "rewards/pad": 0.0, + "step": 189 + }, + { + "completion_length": 97.421875, + "epoch": 0.06054811982154239, + "grad_norm": 138.2771453857422, + "kl": 0.203125, + "learning_rate": 9.394518801784576e-07, + "loss": 0.0081, + "reward": 1.3894553184509277, + "reward_std": 0.11522915214300156, + "rewards/pad": 0.046875, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.34258025884628296, + "step": 190 + }, + { + "completion_length": 123.796875, + "epoch": 0.06086679413639261, + "grad_norm": 100.51148223876953, + "kl": 0.048095703125, + "learning_rate": 9.391332058636074e-07, + "loss": 0.0019, + "reward": 1.4818546772003174, + "reward_std": 0.034188855439424515, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.35685479640960693, + "step": 191 + }, + { + "completion_length": 46.296875, + "epoch": 0.06118546845124283, + "grad_norm": 30.46354866027832, + "kl": 0.126953125, + "learning_rate": 9.388145315487571e-07, + "loss": 0.0051, + "reward": 1.6870028972625732, + "reward_std": 0.13461649417877197, + "rewards/answer_reward": 0.171875, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.515127956867218, + "step": 192 + }, + { + "completion_length": 94.734375, + "epoch": 0.061504142766093055, + "grad_norm": 79.05770111083984, + "kl": 0.1982421875, + "learning_rate": 9.384958572339069e-07, + "loss": 0.0079, + "reward": 1.3827359676361084, + "reward_std": 0.04828489571809769, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3827360272407532, + "step": 193 + }, + { + "completion_length": 46.5, + "epoch": 0.061822817080943275, + "grad_norm": 31.767784118652344, + "kl": 0.2177734375, + "learning_rate": 9.381771829190567e-07, + "loss": 0.0087, + "reward": 1.6183278560638428, + "reward_std": 0.06483359634876251, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4933279752731323, + "rewards/pad": 0.125, + "step": 194 + }, + { + "completion_length": 122.140625, + "epoch": 0.0621414913957935, + "grad_norm": 55.397586822509766, + "kl": 0.10791015625, + "learning_rate": 9.378585086042065e-07, + "loss": 0.0043, + "reward": 1.411421537399292, + "reward_std": 0.042458243668079376, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4114214777946472, + "step": 195 + }, + { + "completion_length": 70.1875, + "epoch": 0.06246016571064372, + "grad_norm": 21.96168327331543, + "kl": 0.24609375, + "learning_rate": 9.375398342893562e-07, + "loss": 0.0098, + "reward": 1.587853193283081, + "reward_std": 0.05139303579926491, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4628531038761139, + "rewards/pad": 0.125, + "step": 196 + }, + { + "completion_length": 73.34375, + "epoch": 0.06277884002549394, + "grad_norm": 138.9192657470703, + "kl": 0.0830078125, + "learning_rate": 9.37221159974506e-07, + "loss": 0.0033, + "reward": 1.7415766716003418, + "reward_std": 0.08817744255065918, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6322016716003418, + "rewards/pad": 0.109375, + "step": 197 + }, + { + "completion_length": 72.1875, + "epoch": 0.06309751434034416, + "grad_norm": 27.619943618774414, + "kl": 0.1103515625, + "learning_rate": 9.369024856596558e-07, + "loss": 0.0044, + "reward": 1.859155535697937, + "reward_std": 0.13565590977668762, + "rewards/answer_reward": 0.40625, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.452905535697937, + "step": 198 + }, + { + "completion_length": 125.484375, + "epoch": 0.0634161886551944, + "grad_norm": 17.163469314575195, + "kl": 0.07080078125, + "learning_rate": 9.365838113448056e-07, + "loss": 0.0028, + "reward": 1.4651894569396973, + "reward_std": 0.04208006337285042, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.34018948674201965, + "step": 199 + }, + { + "completion_length": 96.28125, + "epoch": 0.06373486297004462, + "grad_norm": 76.47855377197266, + "kl": 0.0849609375, + "learning_rate": 9.362651370299553e-07, + "loss": 0.0034, + "reward": 1.6315760612487793, + "reward_std": 0.03873869776725769, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5065759420394897, + "step": 200 + }, + { + "completion_length": 150.71875, + "epoch": 0.06405353728489484, + "grad_norm": 33.26072311401367, + "kl": 0.037109375, + "learning_rate": 9.359464627151051e-07, + "loss": 0.0015, + "reward": 1.3809221982955933, + "reward_std": 0.11550474166870117, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.25592225790023804, + "step": 201 + }, + { + "completion_length": 70.5, + "epoch": 0.06437221159974506, + "grad_norm": 43.25547409057617, + "kl": 0.1689453125, + "learning_rate": 9.356277884002549e-07, + "loss": 0.0067, + "reward": 1.5856120586395264, + "reward_std": 0.0652899295091629, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.46061214804649353, + "rewards/pad": 0.125, + "step": 202 + }, + { + "completion_length": 148.84375, + "epoch": 0.06469088591459528, + "grad_norm": 43.30719757080078, + "kl": 0.060302734375, + "learning_rate": 9.353091140854047e-07, + "loss": 0.0024, + "reward": 1.3250983953475952, + "reward_std": 0.06848461925983429, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3250983655452728, + "rewards/pad": 0.0, + "step": 203 + }, + { + "completion_length": 96.03125, + "epoch": 0.06500956022944551, + "grad_norm": 14.09665584564209, + "kl": 0.220703125, + "learning_rate": 9.349904397705544e-07, + "loss": 0.0088, + "reward": 1.6423754692077637, + "reward_std": 0.05685172975063324, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5173755288124084, + "rewards/pad": 0.125, + "step": 204 + }, + { + "completion_length": 72.328125, + "epoch": 0.06532823454429573, + "grad_norm": 30.157978057861328, + "kl": 0.1015625, + "learning_rate": 9.346717654557042e-07, + "loss": 0.0041, + "reward": 1.7073540687561035, + "reward_std": 0.03902032971382141, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5823540687561035, + "step": 205 + }, + { + "completion_length": 46.40625, + "epoch": 0.06564690885914595, + "grad_norm": 518.1132202148438, + "kl": 0.24609375, + "learning_rate": 9.34353091140854e-07, + "loss": 0.0098, + "reward": 1.7617124319076538, + "reward_std": 0.0738602951169014, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5117124915122986, + "rewards/pad": 0.25, + "step": 206 + }, + { + "completion_length": 45.890625, + "epoch": 0.06596558317399617, + "grad_norm": 57.30687713623047, + "kl": 0.154296875, + "learning_rate": 9.340344168260039e-07, + "loss": 0.0062, + "reward": 1.6535007953643799, + "reward_std": 0.07546480000019073, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5285007953643799, + "step": 207 + }, + { + "completion_length": 72.3125, + "epoch": 0.0662842574888464, + "grad_norm": 134.83474731445312, + "kl": 0.275390625, + "learning_rate": 9.337157425111536e-07, + "loss": 0.011, + "reward": 1.4936914443969727, + "reward_std": 0.08497796952724457, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4936913847923279, + "rewards/pad": 0.0, + "step": 208 + }, + { + "completion_length": 150.125, + "epoch": 0.06660293180369663, + "grad_norm": 63.739009857177734, + "kl": 0.0751953125, + "learning_rate": 9.333970681963034e-07, + "loss": 0.003, + "reward": 1.506711483001709, + "reward_std": 0.03748561441898346, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5067115426063538, + "step": 209 + }, + { + "completion_length": 97.9375, + "epoch": 0.06692160611854685, + "grad_norm": 61.181392669677734, + "kl": 0.08447265625, + "learning_rate": 9.330783938814532e-07, + "loss": 0.0034, + "reward": 1.534330129623413, + "reward_std": 0.09595834463834763, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.42495498061180115, + "rewards/pad": 0.109375, + "step": 210 + }, + { + "completion_length": 72.296875, + "epoch": 0.06724028043339707, + "grad_norm": 139.2161102294922, + "kl": 0.1494140625, + "learning_rate": 9.32759719566603e-07, + "loss": 0.006, + "reward": 1.6207249164581299, + "reward_std": 0.07159844040870667, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4957249164581299, + "rewards/pad": 0.125, + "step": 211 + }, + { + "completion_length": 99.875, + "epoch": 0.06755895474824729, + "grad_norm": 38.422767639160156, + "kl": 0.07763671875, + "learning_rate": 9.324410452517527e-07, + "loss": 0.0031, + "reward": 1.711874008178711, + "reward_std": 0.11063874512910843, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.49312400817871094, + "rewards/pad": 0.21875, + "step": 212 + }, + { + "completion_length": 96.15625, + "epoch": 0.06787762906309751, + "grad_norm": 18.608835220336914, + "kl": 0.1279296875, + "learning_rate": 9.321223709369025e-07, + "loss": 0.0051, + "reward": 1.4139230251312256, + "reward_std": 0.06470437347888947, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4139229357242584, + "rewards/pad": 0.0, + "step": 213 + }, + { + "completion_length": 70.296875, + "epoch": 0.06819630337794774, + "grad_norm": 31.750450134277344, + "kl": 0.134765625, + "learning_rate": 9.318036966220523e-07, + "loss": 0.0054, + "reward": 1.3242194652557373, + "reward_std": 0.058769889175891876, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.32421940565109253, + "rewards/pad": 0.0, + "step": 214 + }, + { + "completion_length": 125.625, + "epoch": 0.06851497769279796, + "grad_norm": 88.57303619384766, + "kl": 0.04736328125, + "learning_rate": 9.314850223072021e-07, + "loss": 0.0019, + "reward": 1.5565061569213867, + "reward_std": 0.17865237593650818, + "rewards/pad": 0.109375, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.4627561569213867, + "step": 215 + }, + { + "completion_length": 99.765625, + "epoch": 0.06883365200764818, + "grad_norm": 37.419498443603516, + "kl": 0.0751953125, + "learning_rate": 9.311663479923517e-07, + "loss": 0.003, + "reward": 1.623964786529541, + "reward_std": 0.15740472078323364, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4520898163318634, + "rewards/pad": 0.171875, + "step": 216 + }, + { + "completion_length": 93.921875, + "epoch": 0.0691523263224984, + "grad_norm": 32.91117858886719, + "kl": 0.09716796875, + "learning_rate": 9.308476736775015e-07, + "loss": 0.0039, + "reward": 1.4654639959335327, + "reward_std": 0.09401823580265045, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.34046393632888794, + "rewards/pad": 0.125, + "step": 217 + }, + { + "completion_length": 98.546875, + "epoch": 0.06947100063734862, + "grad_norm": 32.001346588134766, + "kl": 0.1025390625, + "learning_rate": 9.305289993626513e-07, + "loss": 0.0042, + "reward": 1.5083110332489014, + "reward_std": 0.06683140993118286, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.49268612265586853, + "rewards/pad": 0.015625, + "step": 218 + }, + { + "completion_length": 151.0, + "epoch": 0.06978967495219886, + "grad_norm": 9.118590354919434, + "kl": 0.041015625, + "learning_rate": 9.30210325047801e-07, + "loss": 0.0016, + "reward": 1.685126781463623, + "reward_std": 0.04756775498390198, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.560126781463623, + "step": 219 + }, + { + "completion_length": 71.046875, + "epoch": 0.07010834926704908, + "grad_norm": 16.421247482299805, + "kl": 0.0849609375, + "learning_rate": 9.298916507329508e-07, + "loss": 0.0034, + "reward": 1.5211446285247803, + "reward_std": 0.0653223991394043, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.3961445093154907, + "step": 220 + }, + { + "completion_length": 98.21875, + "epoch": 0.0704270235818993, + "grad_norm": 16.440828323364258, + "kl": 0.08154296875, + "learning_rate": 9.295729764181006e-07, + "loss": 0.0033, + "reward": 1.4838128089904785, + "reward_std": 0.061297886073589325, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4838128089904785, + "step": 221 + }, + { + "completion_length": 73.234375, + "epoch": 0.07074569789674952, + "grad_norm": 138.84011840820312, + "kl": 0.08544921875, + "learning_rate": 9.292543021032504e-07, + "loss": 0.0034, + "reward": 1.5097116231918335, + "reward_std": 0.2135867327451706, + "rewards/pad": 0.09375, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.4315866231918335, + "step": 222 + }, + { + "completion_length": 125.265625, + "epoch": 0.07106437221159974, + "grad_norm": 72.04967498779297, + "kl": 0.0859375, + "learning_rate": 9.289356277884001e-07, + "loss": 0.0034, + "reward": 1.4920474290847778, + "reward_std": 0.07400809228420258, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3670474886894226, + "step": 223 + }, + { + "completion_length": 150.515625, + "epoch": 0.07138304652644997, + "grad_norm": 44.25281524658203, + "kl": 0.06689453125, + "learning_rate": 9.286169534735499e-07, + "loss": 0.0027, + "reward": 1.4957704544067383, + "reward_std": 0.07273541390895844, + "rewards/pad": 0.109375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3863954544067383, + "step": 224 + }, + { + "completion_length": 71.5625, + "epoch": 0.07170172084130019, + "grad_norm": 22.029436111450195, + "kl": 0.09912109375, + "learning_rate": 9.282982791586997e-07, + "loss": 0.004, + "reward": 1.446871280670166, + "reward_std": 0.08580408990383148, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4468713700771332, + "rewards/pad": 0.0, + "step": 225 + }, + { + "completion_length": 122.59375, + "epoch": 0.07202039515615041, + "grad_norm": 50.02418899536133, + "kl": 0.0859375, + "learning_rate": 9.279796048438496e-07, + "loss": 0.0034, + "reward": 1.314070701599121, + "reward_std": 0.024542158469557762, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.31407079100608826, + "step": 226 + }, + { + "completion_length": 98.921875, + "epoch": 0.07233906947100063, + "grad_norm": 26.701772689819336, + "kl": 0.099609375, + "learning_rate": 9.276609305289993e-07, + "loss": 0.004, + "reward": 1.5902478694915771, + "reward_std": 0.14807233214378357, + "rewards/pad": 0.21875, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.37149783968925476, + "step": 227 + }, + { + "completion_length": 121.953125, + "epoch": 0.07265774378585087, + "grad_norm": 51.00769805908203, + "kl": 0.083984375, + "learning_rate": 9.273422562141491e-07, + "loss": 0.0034, + "reward": 1.341005802154541, + "reward_std": 0.06682395935058594, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.341005802154541, + "rewards/pad": 0.0, + "step": 228 + }, + { + "completion_length": 69.484375, + "epoch": 0.07297641810070109, + "grad_norm": 31.971534729003906, + "kl": 0.11767578125, + "learning_rate": 9.270235818992989e-07, + "loss": 0.0047, + "reward": 1.5693368911743164, + "reward_std": 0.07823251187801361, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5693368911743164, + "rewards/pad": 0.0, + "step": 229 + }, + { + "completion_length": 122.03125, + "epoch": 0.07329509241555131, + "grad_norm": 31.826183319091797, + "kl": 0.0673828125, + "learning_rate": 9.267049075844487e-07, + "loss": 0.0027, + "reward": 1.4567553997039795, + "reward_std": 0.035778652876615524, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4567553699016571, + "rewards/pad": 0.0, + "step": 230 + }, + { + "completion_length": 93.734375, + "epoch": 0.07361376673040153, + "grad_norm": 124.7455062866211, + "kl": 0.08251953125, + "learning_rate": 9.263862332695984e-07, + "loss": 0.0033, + "reward": 1.3886239528656006, + "reward_std": 0.06762949377298355, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.38862404227256775, + "rewards/pad": 0.0, + "step": 231 + }, + { + "completion_length": 70.890625, + "epoch": 0.07393244104525175, + "grad_norm": 50.17414093017578, + "kl": 0.283203125, + "learning_rate": 9.260675589547482e-07, + "loss": 0.0114, + "reward": 1.7248307466506958, + "reward_std": 0.08726703375577927, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.599830687046051, + "step": 232 + }, + { + "completion_length": 71.4375, + "epoch": 0.07425111536010198, + "grad_norm": 56.714786529541016, + "kl": 0.13671875, + "learning_rate": 9.25748884639898e-07, + "loss": 0.0055, + "reward": 1.726217269897461, + "reward_std": 0.05578962340950966, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4762171804904938, + "rewards/pad": 0.25, + "step": 233 + }, + { + "completion_length": 18.359375, + "epoch": 0.0745697896749522, + "grad_norm": 53.17715835571289, + "kl": 0.228515625, + "learning_rate": 9.254302103250478e-07, + "loss": 0.0091, + "reward": 1.6254773139953613, + "reward_std": 0.10757525265216827, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6254771947860718, + "rewards/pad": 0.0, + "step": 234 + }, + { + "completion_length": 47.0, + "epoch": 0.07488846398980242, + "grad_norm": 25.17940902709961, + "kl": 0.1826171875, + "learning_rate": 9.251115360101975e-07, + "loss": 0.0073, + "reward": 1.8453280925750732, + "reward_std": 0.15831027925014496, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5015780329704285, + "rewards/pad": 0.34375, + "step": 235 + }, + { + "completion_length": 71.421875, + "epoch": 0.07520713830465264, + "grad_norm": 19.95438003540039, + "kl": 0.1611328125, + "learning_rate": 9.247928616953473e-07, + "loss": 0.0064, + "reward": 1.5744729042053223, + "reward_std": 0.04170050472021103, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.44947290420532227, + "rewards/pad": 0.125, + "step": 236 + }, + { + "completion_length": 74.15625, + "epoch": 0.07552581261950286, + "grad_norm": 28.376657485961914, + "kl": 0.107421875, + "learning_rate": 9.244741873804971e-07, + "loss": 0.0043, + "reward": 1.4814774990081787, + "reward_std": 0.06581274420022964, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3564775884151459, + "step": 237 + }, + { + "completion_length": 72.015625, + "epoch": 0.0758444869343531, + "grad_norm": 34.720123291015625, + "kl": 0.11279296875, + "learning_rate": 9.241555130656469e-07, + "loss": 0.0045, + "reward": 1.7763080596923828, + "reward_std": 0.10871845483779907, + "rewards/answer_reward": 0.234375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5419331192970276, + "step": 238 + }, + { + "completion_length": 69.546875, + "epoch": 0.07616316124920332, + "grad_norm": 125.89852905273438, + "kl": 0.1767578125, + "learning_rate": 9.238368387507966e-07, + "loss": 0.0071, + "reward": 1.3550302982330322, + "reward_std": 0.10732068866491318, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.35503026843070984, + "rewards/pad": 0.0, + "step": 239 + }, + { + "completion_length": 124.578125, + "epoch": 0.07648183556405354, + "grad_norm": 10.14435863494873, + "kl": 0.07275390625, + "learning_rate": 9.235181644359464e-07, + "loss": 0.0029, + "reward": 1.3843462467193604, + "reward_std": 0.03742659091949463, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.38434627652168274, + "step": 240 + }, + { + "completion_length": 47.15625, + "epoch": 0.07680050987890376, + "grad_norm": 72.89696502685547, + "kl": 0.11962890625, + "learning_rate": 9.231994901210962e-07, + "loss": 0.0048, + "reward": 1.5216596126556396, + "reward_std": 0.09107674658298492, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3966595530509949, + "step": 241 + }, + { + "completion_length": 74.703125, + "epoch": 0.07711918419375398, + "grad_norm": 30.012723922729492, + "kl": 0.21875, + "learning_rate": 9.22880815806246e-07, + "loss": 0.0088, + "reward": 1.3861610889434814, + "reward_std": 0.038876067847013474, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.26116111874580383, + "step": 242 + }, + { + "completion_length": 98.4375, + "epoch": 0.07743785850860421, + "grad_norm": 32.56050109863281, + "kl": 0.2421875, + "learning_rate": 9.225621414913957e-07, + "loss": 0.0097, + "reward": 1.386494755744934, + "reward_std": 0.06979185342788696, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3864947557449341, + "step": 243 + }, + { + "completion_length": 71.796875, + "epoch": 0.07775653282345443, + "grad_norm": 24.898000717163086, + "kl": 0.236328125, + "learning_rate": 9.222434671765456e-07, + "loss": 0.0094, + "reward": 1.5179388523101807, + "reward_std": 0.07019379734992981, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5023138523101807, + "rewards/pad": 0.015625, + "step": 244 + }, + { + "completion_length": 98.328125, + "epoch": 0.07807520713830465, + "grad_norm": 44.98802185058594, + "kl": 0.259765625, + "learning_rate": 9.219247928616954e-07, + "loss": 0.0104, + "reward": 1.4779443740844727, + "reward_std": 0.05034922808408737, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.35294443368911743, + "rewards/pad": 0.125, + "step": 245 + }, + { + "completion_length": 47.515625, + "epoch": 0.07839388145315487, + "grad_norm": 58.65922546386719, + "kl": 0.26953125, + "learning_rate": 9.216061185468452e-07, + "loss": 0.0108, + "reward": 1.9427738189697266, + "reward_std": 0.10847378522157669, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5833987593650818, + "rewards/pad": 0.359375, + "step": 246 + }, + { + "completion_length": 97.5625, + "epoch": 0.0787125557680051, + "grad_norm": 48.99842834472656, + "kl": 0.244140625, + "learning_rate": 9.212874442319949e-07, + "loss": 0.0097, + "reward": 1.519799828529358, + "reward_std": 0.06081816181540489, + "rewards/answer_reward": 0.0, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5197998285293579, + "step": 247 + }, + { + "completion_length": 123.5625, + "epoch": 0.07903123008285533, + "grad_norm": 29.7005558013916, + "kl": 0.279296875, + "learning_rate": 9.209687699171447e-07, + "loss": 0.0112, + "reward": 1.5789108276367188, + "reward_std": 0.10013554990291595, + "rewards/pad": 0.109375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.469535768032074, + "step": 248 + }, + { + "completion_length": 124.953125, + "epoch": 0.07934990439770555, + "grad_norm": 57.411460876464844, + "kl": 0.09228515625, + "learning_rate": 9.206500956022945e-07, + "loss": 0.0037, + "reward": 1.602877140045166, + "reward_std": 0.0730300098657608, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.3528771698474884, + "step": 249 + }, + { + "completion_length": 98.1875, + "epoch": 0.07966857871255577, + "grad_norm": 22.777734756469727, + "kl": 0.357421875, + "learning_rate": 9.203314212874442e-07, + "loss": 0.0143, + "reward": 1.448513388633728, + "reward_std": 0.03580144792795181, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.32351335883140564, + "rewards/pad": 0.125, + "step": 250 + }, + { + "completion_length": 70.28125, + "epoch": 0.07998725302740599, + "grad_norm": 53.4788703918457, + "kl": 0.37109375, + "learning_rate": 9.20012746972594e-07, + "loss": 0.0149, + "reward": 1.553734302520752, + "reward_std": 0.13574323058128357, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.33498427271842957, + "rewards/pad": 0.234375, + "step": 251 + }, + { + "completion_length": 152.71875, + "epoch": 0.08030592734225621, + "grad_norm": 20.498197555541992, + "kl": 0.0693359375, + "learning_rate": 9.196940726577438e-07, + "loss": 0.0027, + "reward": 1.369762659072876, + "reward_std": 0.054935187101364136, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.36976274847984314, + "step": 252 + }, + { + "completion_length": 71.046875, + "epoch": 0.08062460165710644, + "grad_norm": 47.005611419677734, + "kl": 0.208984375, + "learning_rate": 9.193753983428936e-07, + "loss": 0.0084, + "reward": 1.6958516836166382, + "reward_std": 0.10709372162818909, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5708516836166382, + "step": 253 + }, + { + "completion_length": 124.765625, + "epoch": 0.08094327597195666, + "grad_norm": 23.856901168823242, + "kl": 0.06884765625, + "learning_rate": 9.190567240280433e-07, + "loss": 0.0028, + "reward": 1.4443206787109375, + "reward_std": 0.03854493796825409, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3193206787109375, + "rewards/pad": 0.125, + "step": 254 + }, + { + "completion_length": 97.53125, + "epoch": 0.08126195028680688, + "grad_norm": 22.07203483581543, + "kl": 0.12451171875, + "learning_rate": 9.18738049713193e-07, + "loss": 0.005, + "reward": 1.414333462715149, + "reward_std": 0.04794395714998245, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.41433337330818176, + "rewards/pad": 0.0, + "step": 255 + }, + { + "completion_length": 71.53125, + "epoch": 0.0815806246016571, + "grad_norm": 33.607948303222656, + "kl": 0.103515625, + "learning_rate": 9.184193753983428e-07, + "loss": 0.0041, + "reward": 1.5641847848892212, + "reward_std": 0.06226586550474167, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4391847550868988, + "rewards/pad": 0.125, + "step": 256 + }, + { + "completion_length": 123.53125, + "epoch": 0.08189929891650732, + "grad_norm": 11.251544952392578, + "kl": 0.1015625, + "learning_rate": 9.181007010834926e-07, + "loss": 0.0041, + "reward": 1.3956583738327026, + "reward_std": 0.10355216264724731, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.41128334403038025, + "step": 257 + }, + { + "completion_length": 123.890625, + "epoch": 0.08221797323135756, + "grad_norm": 32.35093307495117, + "kl": 0.07421875, + "learning_rate": 9.177820267686423e-07, + "loss": 0.003, + "reward": 1.4286943674087524, + "reward_std": 0.04091978818178177, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4286944270133972, + "step": 258 + }, + { + "completion_length": 73.65625, + "epoch": 0.08253664754620778, + "grad_norm": 42.930885314941406, + "kl": 0.138671875, + "learning_rate": 9.174633524537921e-07, + "loss": 0.0055, + "reward": 1.6769955158233643, + "reward_std": 0.13682706654071808, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5207455158233643, + "rewards/pad": 0.15625, + "step": 259 + }, + { + "completion_length": 45.46875, + "epoch": 0.082855321861058, + "grad_norm": 73.59136199951172, + "kl": 0.5703125, + "learning_rate": 9.171446781389419e-07, + "loss": 0.0229, + "reward": 1.7261182069778442, + "reward_std": 0.13216665387153625, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4761181175708771, + "step": 260 + }, + { + "completion_length": 99.265625, + "epoch": 0.08317399617590822, + "grad_norm": 50.60124969482422, + "kl": 0.123046875, + "learning_rate": 9.168260038240917e-07, + "loss": 0.0049, + "reward": 1.5830243825912476, + "reward_std": 0.06387736648321152, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.47364941239356995, + "rewards/pad": 0.109375, + "step": 261 + }, + { + "completion_length": 95.359375, + "epoch": 0.08349267049075844, + "grad_norm": 40.418121337890625, + "kl": 0.1484375, + "learning_rate": 9.165073295092414e-07, + "loss": 0.0059, + "reward": 1.3880648612976074, + "reward_std": 0.09246179461479187, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3880648612976074, + "step": 262 + }, + { + "completion_length": 146.625, + "epoch": 0.08381134480560867, + "grad_norm": 34.34090805053711, + "kl": 0.0859375, + "learning_rate": 9.161886551943912e-07, + "loss": 0.0034, + "reward": 1.482377052307129, + "reward_std": 0.03902214393019676, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.48237714171409607, + "step": 263 + }, + { + "completion_length": 124.390625, + "epoch": 0.0841300191204589, + "grad_norm": 19.643157958984375, + "kl": 0.11572265625, + "learning_rate": 9.158699808795411e-07, + "loss": 0.0046, + "reward": 1.3950953483581543, + "reward_std": 0.0343056246638298, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3950953483581543, + "rewards/pad": 0.0, + "step": 264 + }, + { + "completion_length": 45.71875, + "epoch": 0.08444869343530911, + "grad_norm": 109.31044006347656, + "kl": 0.216796875, + "learning_rate": 9.155513065646909e-07, + "loss": 0.0086, + "reward": 1.6760146617889404, + "reward_std": 0.10303567349910736, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5510146021842957, + "step": 265 + }, + { + "completion_length": 71.390625, + "epoch": 0.08476736775015933, + "grad_norm": 78.38106536865234, + "kl": 0.1689453125, + "learning_rate": 9.152326322498406e-07, + "loss": 0.0068, + "reward": 1.678178071975708, + "reward_std": 0.0742669403553009, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.553178071975708, + "rewards/pad": 0.125, + "step": 266 + }, + { + "completion_length": 121.96875, + "epoch": 0.08508604206500955, + "grad_norm": 56.864837646484375, + "kl": 0.095703125, + "learning_rate": 9.149139579349904e-07, + "loss": 0.0038, + "reward": 1.3991233110427856, + "reward_std": 0.035200487822294235, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3991233706474304, + "step": 267 + }, + { + "completion_length": 71.0625, + "epoch": 0.08540471637985979, + "grad_norm": 30.662776947021484, + "kl": 0.1484375, + "learning_rate": 9.145952836201402e-07, + "loss": 0.0059, + "reward": 1.4969425201416016, + "reward_std": 0.13425540924072266, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.38756754994392395, + "rewards/pad": 0.125, + "step": 268 + }, + { + "completion_length": 97.734375, + "epoch": 0.08572339069471001, + "grad_norm": 57.5777587890625, + "kl": 0.138671875, + "learning_rate": 9.1427660930529e-07, + "loss": 0.0056, + "reward": 1.4289636611938477, + "reward_std": 0.05716238170862198, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.42896372079849243, + "rewards/pad": 0.0, + "step": 269 + }, + { + "completion_length": 175.0625, + "epoch": 0.08604206500956023, + "grad_norm": 39.821189880371094, + "kl": 0.06982421875, + "learning_rate": 9.139579349904397e-07, + "loss": 0.0028, + "reward": 1.4713307619094849, + "reward_std": 0.03898857533931732, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.47133076190948486, + "step": 270 + }, + { + "completion_length": 102.734375, + "epoch": 0.08636073932441045, + "grad_norm": 37.10630416870117, + "kl": 0.1181640625, + "learning_rate": 9.136392606755895e-07, + "loss": 0.0047, + "reward": 1.4957199096679688, + "reward_std": 0.05200393870472908, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.37071990966796875, + "step": 271 + }, + { + "completion_length": 73.59375, + "epoch": 0.08667941363926067, + "grad_norm": 135.67630004882812, + "kl": 0.11865234375, + "learning_rate": 9.133205863607393e-07, + "loss": 0.0047, + "reward": 1.598374843597412, + "reward_std": 0.15692463517189026, + "rewards/answer_reward": 0.0625, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5358746647834778, + "step": 272 + }, + { + "completion_length": 71.78125, + "epoch": 0.0869980879541109, + "grad_norm": 41.58750915527344, + "kl": 0.2001953125, + "learning_rate": 9.130019120458891e-07, + "loss": 0.008, + "reward": 1.4793663024902344, + "reward_std": 0.07295481860637665, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4793663024902344, + "rewards/pad": 0.0, + "step": 273 + }, + { + "completion_length": 44.828125, + "epoch": 0.08731676226896112, + "grad_norm": 62.157081604003906, + "kl": 0.138671875, + "learning_rate": 9.126832377310388e-07, + "loss": 0.0056, + "reward": 1.5699678659439087, + "reward_std": 0.09194605052471161, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4918428659439087, + "rewards/pad": 0.078125, + "step": 274 + }, + { + "completion_length": 96.859375, + "epoch": 0.08763543658381134, + "grad_norm": 34.54340744018555, + "kl": 0.298828125, + "learning_rate": 9.123645634161886e-07, + "loss": 0.0119, + "reward": 1.5665819644927979, + "reward_std": 0.07807460427284241, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.44158199429512024, + "step": 275 + }, + { + "completion_length": 96.1875, + "epoch": 0.08795411089866156, + "grad_norm": 72.8243179321289, + "kl": 0.09521484375, + "learning_rate": 9.120458891013384e-07, + "loss": 0.0038, + "reward": 1.6681749820709229, + "reward_std": 0.051725488156080246, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5431750416755676, + "rewards/pad": 0.125, + "step": 276 + }, + { + "completion_length": 122.734375, + "epoch": 0.08827278521351178, + "grad_norm": 25.338285446166992, + "kl": 0.361328125, + "learning_rate": 9.117272147864882e-07, + "loss": 0.0144, + "reward": 1.3991707563400269, + "reward_std": 0.07661792635917664, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.39917081594467163, + "rewards/pad": 0.0, + "step": 277 + }, + { + "completion_length": 20.0625, + "epoch": 0.08859145952836202, + "grad_norm": 56.99957275390625, + "kl": 0.1826171875, + "learning_rate": 9.114085404716379e-07, + "loss": 0.0073, + "reward": 1.7226518392562866, + "reward_std": 0.11980736255645752, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6132767796516418, + "rewards/pad": 0.109375, + "step": 278 + }, + { + "completion_length": 124.9375, + "epoch": 0.08891013384321224, + "grad_norm": 16.744503021240234, + "kl": 0.0654296875, + "learning_rate": 9.110898661567877e-07, + "loss": 0.0026, + "reward": 1.5589927434921265, + "reward_std": 0.042084500193595886, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4339926838874817, + "step": 279 + }, + { + "completion_length": 95.1875, + "epoch": 0.08922880815806246, + "grad_norm": 39.232322692871094, + "kl": 0.08740234375, + "learning_rate": 9.107711918419375e-07, + "loss": 0.0035, + "reward": 1.452876329421997, + "reward_std": 0.17438113689422607, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.46850135922431946, + "step": 280 + }, + { + "completion_length": 123.921875, + "epoch": 0.08954748247291268, + "grad_norm": 31.64417839050293, + "kl": 0.12353515625, + "learning_rate": 9.104525175270872e-07, + "loss": 0.005, + "reward": 1.5177637338638306, + "reward_std": 0.055521003901958466, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5177636742591858, + "step": 281 + }, + { + "completion_length": 19.5, + "epoch": 0.08986615678776291, + "grad_norm": 32.218997955322266, + "kl": 0.29296875, + "learning_rate": 9.10133843212237e-07, + "loss": 0.0117, + "reward": 1.470993995666504, + "reward_std": 0.12481606006622314, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4397439956665039, + "rewards/pad": 0.03125, + "step": 282 + }, + { + "completion_length": 46.0625, + "epoch": 0.09018483110261313, + "grad_norm": 43.10451126098633, + "kl": 0.1357421875, + "learning_rate": 9.098151688973869e-07, + "loss": 0.0054, + "reward": 1.586465835571289, + "reward_std": 0.15732800960540771, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4302159249782562, + "rewards/pad": 0.15625, + "step": 283 + }, + { + "completion_length": 97.484375, + "epoch": 0.09050350541746335, + "grad_norm": 19.141231536865234, + "kl": 0.09228515625, + "learning_rate": 9.094964945825367e-07, + "loss": 0.0037, + "reward": 1.7354822158813477, + "reward_std": 0.07660651952028275, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5011070966720581, + "rewards/pad": 0.234375, + "step": 284 + }, + { + "completion_length": 98.140625, + "epoch": 0.09082217973231357, + "grad_norm": 14.528425216674805, + "kl": 0.09765625, + "learning_rate": 9.091778202676864e-07, + "loss": 0.0039, + "reward": 1.5818281173706055, + "reward_std": 0.05947490036487579, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.45682817697525024, + "rewards/pad": 0.125, + "step": 285 + }, + { + "completion_length": 98.359375, + "epoch": 0.0911408540471638, + "grad_norm": 15.328421592712402, + "kl": 0.07861328125, + "learning_rate": 9.088591459528362e-07, + "loss": 0.0031, + "reward": 1.688610553741455, + "reward_std": 0.05649499222636223, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4386104941368103, + "step": 286 + }, + { + "completion_length": 124.65625, + "epoch": 0.09145952836201403, + "grad_norm": 10.835719108581543, + "kl": 0.0810546875, + "learning_rate": 9.08540471637986e-07, + "loss": 0.0032, + "reward": 1.665027141571045, + "reward_std": 0.10295233130455017, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.5556520819664001, + "step": 287 + }, + { + "completion_length": 150.796875, + "epoch": 0.09177820267686425, + "grad_norm": 14.098941802978516, + "kl": 0.07568359375, + "learning_rate": 9.082217973231358e-07, + "loss": 0.0031, + "reward": 1.432703971862793, + "reward_std": 0.05922875553369522, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.43270397186279297, + "step": 288 + }, + { + "completion_length": 46.125, + "epoch": 0.09209687699171447, + "grad_norm": 80.2524185180664, + "kl": 0.216796875, + "learning_rate": 9.079031230082855e-07, + "loss": 0.0087, + "reward": 1.7530198097229004, + "reward_std": 0.10651280730962753, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6280196905136108, + "rewards/pad": 0.125, + "step": 289 + }, + { + "completion_length": 97.125, + "epoch": 0.09241555130656469, + "grad_norm": 15.310797691345215, + "kl": 0.166015625, + "learning_rate": 9.075844486934353e-07, + "loss": 0.0066, + "reward": 1.7720351219177246, + "reward_std": 0.13050448894500732, + "rewards/pad": 0.234375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5376600623130798, + "step": 290 + }, + { + "completion_length": 121.671875, + "epoch": 0.09273422562141491, + "grad_norm": 15.590791702270508, + "kl": 0.09033203125, + "learning_rate": 9.072657743785851e-07, + "loss": 0.0036, + "reward": 1.4074218273162842, + "reward_std": 0.04259195923805237, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4074217677116394, + "step": 291 + }, + { + "completion_length": 18.265625, + "epoch": 0.09305289993626514, + "grad_norm": 36.518375396728516, + "kl": 0.55859375, + "learning_rate": 9.069471000637349e-07, + "loss": 0.0224, + "reward": 1.3181414604187012, + "reward_std": 0.08523581176996231, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.31814149022102356, + "rewards/pad": 0.0, + "step": 292 + }, + { + "completion_length": 97.609375, + "epoch": 0.09337157425111536, + "grad_norm": 25.183494567871094, + "kl": 0.08984375, + "learning_rate": 9.066284257488846e-07, + "loss": 0.0036, + "reward": 1.4201927185058594, + "reward_std": 0.04629743844270706, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4201928973197937, + "step": 293 + }, + { + "completion_length": 72.8125, + "epoch": 0.09369024856596558, + "grad_norm": 23.733295440673828, + "kl": 0.24609375, + "learning_rate": 9.063097514340344e-07, + "loss": 0.0098, + "reward": 1.611457347869873, + "reward_std": 0.08153215050697327, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.48645731806755066, + "rewards/pad": 0.125, + "step": 294 + }, + { + "completion_length": 99.671875, + "epoch": 0.0940089228808158, + "grad_norm": 27.60891342163086, + "kl": 0.2109375, + "learning_rate": 9.059910771191841e-07, + "loss": 0.0084, + "reward": 1.6462310552597046, + "reward_std": 0.08268588036298752, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.5368560552597046, + "rewards/pad": 0.125, + "step": 295 + }, + { + "completion_length": 97.9375, + "epoch": 0.09432759719566602, + "grad_norm": 60.23105239868164, + "kl": 0.1171875, + "learning_rate": 9.056724028043339e-07, + "loss": 0.0047, + "reward": 1.507036805152893, + "reward_std": 0.24251246452331543, + "rewards/pad": 0.0625, + "rewards/tracking_format_reward": 0.96875, + "rewards/tracking_iou_reward": 0.4757867455482483, + "step": 296 + }, + { + "completion_length": 122.328125, + "epoch": 0.09464627151051626, + "grad_norm": 31.52186393737793, + "kl": 0.0869140625, + "learning_rate": 9.053537284894836e-07, + "loss": 0.0035, + "reward": 1.488356113433838, + "reward_std": 0.062449708580970764, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4883561432361603, + "rewards/pad": 0.0, + "step": 297 + }, + { + "completion_length": 97.96875, + "epoch": 0.09496494582536648, + "grad_norm": 13.065885543823242, + "kl": 0.10498046875, + "learning_rate": 9.050350541746334e-07, + "loss": 0.0042, + "reward": 1.5660672187805176, + "reward_std": 0.03638705611228943, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4410672187805176, + "rewards/pad": 0.125, + "step": 298 + }, + { + "completion_length": 96.75, + "epoch": 0.0952836201402167, + "grad_norm": 38.61499786376953, + "kl": 0.10107421875, + "learning_rate": 9.047163798597832e-07, + "loss": 0.004, + "reward": 1.4196743965148926, + "reward_std": 0.06692565977573395, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4196743071079254, + "rewards/pad": 0.0, + "step": 299 + }, + { + "completion_length": 99.8125, + "epoch": 0.09560229445506692, + "grad_norm": 31.40985679626465, + "kl": 0.07177734375, + "learning_rate": 9.04397705544933e-07, + "loss": 0.0029, + "reward": 1.7245066165924072, + "reward_std": 0.05942245572805405, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4745066165924072, + "step": 300 + }, + { + "completion_length": 99.34375, + "epoch": 0.09592096876991714, + "grad_norm": 13.066228866577148, + "kl": 0.1318359375, + "learning_rate": 9.040790312300827e-07, + "loss": 0.0053, + "reward": 1.365534782409668, + "reward_std": 0.09436441212892532, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.38115984201431274, + "rewards/pad": 0.0, + "step": 301 + }, + { + "completion_length": 97.21875, + "epoch": 0.09623964308476737, + "grad_norm": 45.45714569091797, + "kl": 0.083984375, + "learning_rate": 9.037603569152326e-07, + "loss": 0.0034, + "reward": 1.6654975414276123, + "reward_std": 0.15218521654605865, + "rewards/answer_reward": 0.203125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.46237245202064514, + "step": 302 + }, + { + "completion_length": 123.859375, + "epoch": 0.0965583173996176, + "grad_norm": 62.40977096557617, + "kl": 0.08349609375, + "learning_rate": 9.034416826003824e-07, + "loss": 0.0033, + "reward": 1.310499668121338, + "reward_std": 0.06527701765298843, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3104996085166931, + "step": 303 + }, + { + "completion_length": 70.203125, + "epoch": 0.09687699171446781, + "grad_norm": 78.83379364013672, + "kl": 0.080078125, + "learning_rate": 9.031230082855322e-07, + "loss": 0.0032, + "reward": 1.51714289188385, + "reward_std": 0.09799807518720627, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5171428918838501, + "step": 304 + }, + { + "completion_length": 148.671875, + "epoch": 0.09719566602931803, + "grad_norm": 7.507280349731445, + "kl": 0.05908203125, + "learning_rate": 9.028043339706819e-07, + "loss": 0.0024, + "reward": 1.4250938892364502, + "reward_std": 0.02290337160229683, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.42509400844573975, + "rewards/pad": 0.0, + "step": 305 + }, + { + "completion_length": 44.546875, + "epoch": 0.09751434034416825, + "grad_norm": 109.25292205810547, + "kl": 0.1220703125, + "learning_rate": 9.024856596558317e-07, + "loss": 0.0049, + "reward": 1.4947134256362915, + "reward_std": 0.08180755376815796, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.49471336603164673, + "rewards/pad": 0.0, + "step": 306 + }, + { + "completion_length": 48.046875, + "epoch": 0.09783301465901849, + "grad_norm": 135.5470428466797, + "kl": 0.125, + "learning_rate": 9.021669853409815e-07, + "loss": 0.005, + "reward": 1.6899664402008057, + "reward_std": 0.14552678167819977, + "rewards/answer_reward": 0.3125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.3774664103984833, + "step": 307 + }, + { + "completion_length": 71.75, + "epoch": 0.09815168897386871, + "grad_norm": 19.311742782592773, + "kl": 0.1279296875, + "learning_rate": 9.018483110261312e-07, + "loss": 0.0051, + "reward": 1.4897065162658691, + "reward_std": 0.054242588579654694, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.48970645666122437, + "rewards/pad": 0.0, + "step": 308 + }, + { + "completion_length": 97.0625, + "epoch": 0.09847036328871893, + "grad_norm": 14.59001636505127, + "kl": 0.1982421875, + "learning_rate": 9.01529636711281e-07, + "loss": 0.0079, + "reward": 1.43793523311615, + "reward_std": 0.05904996767640114, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4379352629184723, + "step": 309 + }, + { + "completion_length": 123.28125, + "epoch": 0.09878903760356915, + "grad_norm": 42.15071487426758, + "kl": 0.039794921875, + "learning_rate": 9.012109623964308e-07, + "loss": 0.0016, + "reward": 1.7498985528945923, + "reward_std": 0.04076165705919266, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4998985528945923, + "step": 310 + }, + { + "completion_length": 74.34375, + "epoch": 0.09910771191841937, + "grad_norm": 77.68610382080078, + "kl": 0.126953125, + "learning_rate": 9.008922880815806e-07, + "loss": 0.0051, + "reward": 1.4720065593719482, + "reward_std": 0.10089914500713348, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4720064103603363, + "rewards/pad": 0.0, + "step": 311 + }, + { + "completion_length": 176.0625, + "epoch": 0.0994263862332696, + "grad_norm": 46.3254508972168, + "kl": 0.036376953125, + "learning_rate": 9.005736137667303e-07, + "loss": 0.0015, + "reward": 1.5705065727233887, + "reward_std": 0.030933646485209465, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.44550663232803345, + "step": 312 + }, + { + "completion_length": 71.703125, + "epoch": 0.09974506054811982, + "grad_norm": 50.289485931396484, + "kl": 0.380859375, + "learning_rate": 9.002549394518801e-07, + "loss": 0.0153, + "reward": 1.5453498363494873, + "reward_std": 0.053975529968738556, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.42034971714019775, + "step": 313 + }, + { + "completion_length": 97.25, + "epoch": 0.10006373486297004, + "grad_norm": 96.94417572021484, + "kl": 0.8125, + "learning_rate": 8.999362651370299e-07, + "loss": 0.0326, + "reward": 1.7055671215057373, + "reward_std": 0.07332450896501541, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5805671215057373, + "step": 314 + }, + { + "completion_length": 95.4375, + "epoch": 0.10038240917782026, + "grad_norm": 81.17061614990234, + "kl": 0.07861328125, + "learning_rate": 8.996175908221797e-07, + "loss": 0.0031, + "reward": 1.5590391159057617, + "reward_std": 0.14300386607646942, + "rewards/pad": 0.078125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.48091405630111694, + "step": 315 + }, + { + "completion_length": 72.828125, + "epoch": 0.10070108349267048, + "grad_norm": 88.43734741210938, + "kl": 0.79296875, + "learning_rate": 8.992989165073294e-07, + "loss": 0.0318, + "reward": 1.6209162473678589, + "reward_std": 0.1531776785850525, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4959162473678589, + "rewards/pad": 0.125, + "step": 316 + }, + { + "completion_length": 68.109375, + "epoch": 0.10101975780752072, + "grad_norm": 31.90052032470703, + "kl": 0.12890625, + "learning_rate": 8.989802421924792e-07, + "loss": 0.0052, + "reward": 1.5254950523376465, + "reward_std": 0.07256342470645905, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5254950523376465, + "rewards/pad": 0.0, + "step": 317 + }, + { + "completion_length": 71.765625, + "epoch": 0.10133843212237094, + "grad_norm": 27.011852264404297, + "kl": 0.25390625, + "learning_rate": 8.98661567877629e-07, + "loss": 0.0101, + "reward": 1.567828893661499, + "reward_std": 0.12586727738380432, + "rewards/answer_reward": 0.0, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.567828893661499, + "step": 318 + }, + { + "completion_length": 124.890625, + "epoch": 0.10165710643722116, + "grad_norm": 28.799516677856445, + "kl": 0.09716796875, + "learning_rate": 8.983428935627788e-07, + "loss": 0.0039, + "reward": 1.7499445676803589, + "reward_std": 0.04650557041168213, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4999445080757141, + "step": 319 + }, + { + "completion_length": 98.84375, + "epoch": 0.10197578075207138, + "grad_norm": 22.20235252380371, + "kl": 0.10400390625, + "learning_rate": 8.980242192479286e-07, + "loss": 0.0041, + "reward": 1.466164231300354, + "reward_std": 0.0942983478307724, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4661642909049988, + "step": 320 + }, + { + "completion_length": 98.28125, + "epoch": 0.1022944550669216, + "grad_norm": 35.46802520751953, + "kl": 0.1015625, + "learning_rate": 8.977055449330784e-07, + "loss": 0.004, + "reward": 1.5338505506515503, + "reward_std": 0.050848715007305145, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.2838504910469055, + "step": 321 + }, + { + "completion_length": 76.0, + "epoch": 0.10261312938177183, + "grad_norm": 28.48044776916504, + "kl": 0.1181640625, + "learning_rate": 8.973868706182282e-07, + "loss": 0.0047, + "reward": 1.3551535606384277, + "reward_std": 0.09504449367523193, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3551536798477173, + "rewards/pad": 0.0, + "step": 322 + }, + { + "completion_length": 45.03125, + "epoch": 0.10293180369662205, + "grad_norm": 20.255050659179688, + "kl": 0.1435546875, + "learning_rate": 8.97068196303378e-07, + "loss": 0.0058, + "reward": 1.4209308624267578, + "reward_std": 0.0825200229883194, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.42093077301979065, + "rewards/pad": 0.0, + "step": 323 + }, + { + "completion_length": 127.28125, + "epoch": 0.10325047801147227, + "grad_norm": 31.418140411376953, + "kl": 0.06640625, + "learning_rate": 8.967495219885277e-07, + "loss": 0.0027, + "reward": 1.5614705085754395, + "reward_std": 0.08935748040676117, + "rewards/pad": 0.140625, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4208453893661499, + "step": 324 + }, + { + "completion_length": 70.859375, + "epoch": 0.1035691523263225, + "grad_norm": 46.53847122192383, + "kl": 0.09814453125, + "learning_rate": 8.964308476736775e-07, + "loss": 0.0039, + "reward": 1.7592271566390991, + "reward_std": 0.10191182047128677, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.3998522162437439, + "rewards/pad": 0.375, + "step": 325 + }, + { + "completion_length": 45.0625, + "epoch": 0.10388782664117271, + "grad_norm": 63.19196701049805, + "kl": 0.1142578125, + "learning_rate": 8.961121733588273e-07, + "loss": 0.0046, + "reward": 1.588505506515503, + "reward_std": 0.10087063908576965, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4635055661201477, + "rewards/pad": 0.125, + "step": 326 + }, + { + "completion_length": 177.09375, + "epoch": 0.10420650095602295, + "grad_norm": 14.488212585449219, + "kl": 0.032958984375, + "learning_rate": 8.957934990439771e-07, + "loss": 0.0013, + "reward": 1.3769497871398926, + "reward_std": 0.03301223739981651, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3769497275352478, + "step": 327 + }, + { + "completion_length": 72.4375, + "epoch": 0.10452517527087317, + "grad_norm": 28.563243865966797, + "kl": 0.1337890625, + "learning_rate": 8.954748247291268e-07, + "loss": 0.0053, + "reward": 1.5478661060333252, + "reward_std": 0.12154059112071991, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3603661060333252, + "rewards/pad": 0.1875, + "step": 328 + }, + { + "completion_length": 125.125, + "epoch": 0.10484384958572339, + "grad_norm": 199.4341583251953, + "kl": 0.07958984375, + "learning_rate": 8.951561504142766e-07, + "loss": 0.0032, + "reward": 1.3509695529937744, + "reward_std": 0.12035353481769562, + "rewards/answer_reward": 0.140625, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.21034449338912964, + "step": 329 + }, + { + "completion_length": 18.8125, + "epoch": 0.10516252390057361, + "grad_norm": 626.1193237304688, + "kl": 0.197265625, + "learning_rate": 8.948374760994264e-07, + "loss": 0.0079, + "reward": 1.6512441635131836, + "reward_std": 0.1490609496831894, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6512441635131836, + "rewards/pad": 0.0, + "step": 330 + }, + { + "completion_length": 122.71875, + "epoch": 0.10548119821542384, + "grad_norm": 24.436813354492188, + "kl": 0.1044921875, + "learning_rate": 8.945188017845762e-07, + "loss": 0.0042, + "reward": 1.5120811462402344, + "reward_std": 0.03990060091018677, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5120812058448792, + "step": 331 + }, + { + "completion_length": 73.984375, + "epoch": 0.10579987253027406, + "grad_norm": 24.41294288635254, + "kl": 0.130859375, + "learning_rate": 8.942001274697259e-07, + "loss": 0.0052, + "reward": 1.5127191543579102, + "reward_std": 0.06821781396865845, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3877192735671997, + "rewards/pad": 0.125, + "step": 332 + }, + { + "completion_length": 121.734375, + "epoch": 0.10611854684512428, + "grad_norm": 22.645463943481445, + "kl": 0.0732421875, + "learning_rate": 8.938814531548757e-07, + "loss": 0.003, + "reward": 1.3383376598358154, + "reward_std": 0.027271658182144165, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.33833763003349304, + "rewards/pad": 0.0, + "step": 333 + }, + { + "completion_length": 149.6875, + "epoch": 0.1064372211599745, + "grad_norm": 15.65821361541748, + "kl": 0.039794921875, + "learning_rate": 8.935627788400254e-07, + "loss": 0.0016, + "reward": 1.5990335941314697, + "reward_std": 0.04202403873205185, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4740336537361145, + "step": 334 + }, + { + "completion_length": 178.015625, + "epoch": 0.10675589547482472, + "grad_norm": 25.23544692993164, + "kl": 0.0439453125, + "learning_rate": 8.932441045251752e-07, + "loss": 0.0018, + "reward": 1.4337772130966187, + "reward_std": 0.08441469818353653, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.4494021236896515, + "step": 335 + }, + { + "completion_length": 144.8125, + "epoch": 0.10707456978967496, + "grad_norm": 22.692462921142578, + "kl": 0.0703125, + "learning_rate": 8.929254302103249e-07, + "loss": 0.0028, + "reward": 1.4097086191177368, + "reward_std": 0.05583653971552849, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.40970855951309204, + "step": 336 + }, + { + "completion_length": 170.28125, + "epoch": 0.10739324410452518, + "grad_norm": 44.01387405395508, + "kl": 0.038818359375, + "learning_rate": 8.926067558954747e-07, + "loss": 0.0016, + "reward": 1.407758116722107, + "reward_std": 0.05321196839213371, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.28275805711746216, + "step": 337 + }, + { + "completion_length": 98.15625, + "epoch": 0.1077119184193754, + "grad_norm": 72.85901641845703, + "kl": 0.0966796875, + "learning_rate": 8.922880815806245e-07, + "loss": 0.0039, + "reward": 1.5778709650039673, + "reward_std": 0.08505690097808838, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.45287102460861206, + "rewards/pad": 0.125, + "step": 338 + }, + { + "completion_length": 98.4375, + "epoch": 0.10803059273422562, + "grad_norm": 18.919353485107422, + "kl": 0.0966796875, + "learning_rate": 8.919694072657742e-07, + "loss": 0.0039, + "reward": 1.6748125553131104, + "reward_std": 0.04199599474668503, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5498126149177551, + "rewards/pad": 0.125, + "step": 339 + }, + { + "completion_length": 95.484375, + "epoch": 0.10834926704907584, + "grad_norm": 17.899627685546875, + "kl": 0.1201171875, + "learning_rate": 8.916507329509241e-07, + "loss": 0.0048, + "reward": 1.4470590353012085, + "reward_std": 0.06309215724468231, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3220589756965637, + "rewards/pad": 0.125, + "step": 340 + }, + { + "completion_length": 45.546875, + "epoch": 0.10866794136392607, + "grad_norm": 34.16611099243164, + "kl": 0.1611328125, + "learning_rate": 8.913320586360739e-07, + "loss": 0.0064, + "reward": 1.5854992866516113, + "reward_std": 0.10462392866611481, + "rewards/pad": 0.09375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.49174928665161133, + "step": 341 + }, + { + "completion_length": 73.15625, + "epoch": 0.1089866156787763, + "grad_norm": 30.530794143676758, + "kl": 0.177734375, + "learning_rate": 8.910133843212237e-07, + "loss": 0.0071, + "reward": 1.880784511566162, + "reward_std": 0.06079376861453056, + "rewards/answer_reward": 0.375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5057844519615173, + "step": 342 + }, + { + "completion_length": 72.015625, + "epoch": 0.10930528999362651, + "grad_norm": 19.3212833404541, + "kl": 0.142578125, + "learning_rate": 8.906947100063734e-07, + "loss": 0.0057, + "reward": 1.5231934785842896, + "reward_std": 0.05362851545214653, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.39819350838661194, + "rewards/pad": 0.125, + "step": 343 + }, + { + "completion_length": 44.890625, + "epoch": 0.10962396430847673, + "grad_norm": 113.36170196533203, + "kl": 0.671875, + "learning_rate": 8.903760356915232e-07, + "loss": 0.0269, + "reward": 1.7440569400787354, + "reward_std": 0.13982772827148438, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.6190569400787354, + "step": 344 + }, + { + "completion_length": 94.1875, + "epoch": 0.10994263862332695, + "grad_norm": 84.85638427734375, + "kl": 0.10986328125, + "learning_rate": 8.90057361376673e-07, + "loss": 0.0044, + "reward": 1.4598510265350342, + "reward_std": 0.09140007197856903, + "rewards/answer_reward": 0.109375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.3504760265350342, + "step": 345 + }, + { + "completion_length": 124.046875, + "epoch": 0.11026131293817719, + "grad_norm": 106.96361541748047, + "kl": 0.10595703125, + "learning_rate": 8.897386870618228e-07, + "loss": 0.0042, + "reward": 1.5757009983062744, + "reward_std": 0.11959008872509003, + "rewards/pad": 0.0625, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5132008790969849, + "step": 346 + }, + { + "completion_length": 98.046875, + "epoch": 0.11057998725302741, + "grad_norm": 62.11159133911133, + "kl": 0.404296875, + "learning_rate": 8.894200127469725e-07, + "loss": 0.0162, + "reward": 1.4973745346069336, + "reward_std": 0.09182165563106537, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.35674965381622314, + "rewards/pad": 0.140625, + "step": 347 + }, + { + "completion_length": 70.578125, + "epoch": 0.11089866156787763, + "grad_norm": 28.104806900024414, + "kl": 0.1357421875, + "learning_rate": 8.891013384321223e-07, + "loss": 0.0054, + "reward": 1.7881293296813965, + "reward_std": 0.10459479689598083, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.538129448890686, + "step": 348 + }, + { + "completion_length": 124.671875, + "epoch": 0.11121733588272785, + "grad_norm": 39.63365936279297, + "kl": 0.08544921875, + "learning_rate": 8.887826641172721e-07, + "loss": 0.0034, + "reward": 1.4937341213226318, + "reward_std": 0.04106447100639343, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.49373406171798706, + "rewards/pad": 0.0, + "step": 349 + }, + { + "completion_length": 96.65625, + "epoch": 0.11153601019757807, + "grad_norm": 13.17743968963623, + "kl": 0.115234375, + "learning_rate": 8.884639898024219e-07, + "loss": 0.0046, + "reward": 1.3785258531570435, + "reward_std": 0.05048896372318268, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.37852585315704346, + "rewards/pad": 0.0, + "step": 350 + }, + { + "completion_length": 150.3125, + "epoch": 0.1118546845124283, + "grad_norm": 10.41886043548584, + "kl": 0.0703125, + "learning_rate": 8.881453154875716e-07, + "loss": 0.0029, + "reward": 1.6018152236938477, + "reward_std": 0.023036912083625793, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.47681528329849243, + "step": 351 + }, + { + "completion_length": 98.09375, + "epoch": 0.11217335882727852, + "grad_norm": 24.771432876586914, + "kl": 0.1845703125, + "learning_rate": 8.878266411727214e-07, + "loss": 0.0074, + "reward": 1.751751184463501, + "reward_std": 0.07713186740875244, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.501751184463501, + "step": 352 + }, + { + "completion_length": 95.296875, + "epoch": 0.11249203314212874, + "grad_norm": 108.3392333984375, + "kl": 0.09619140625, + "learning_rate": 8.875079668578712e-07, + "loss": 0.0038, + "reward": 1.344058871269226, + "reward_std": 0.06936553120613098, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.34405890107154846, + "step": 353 + }, + { + "completion_length": 121.9375, + "epoch": 0.11281070745697896, + "grad_norm": 16.075292587280273, + "kl": 0.1689453125, + "learning_rate": 8.87189292543021e-07, + "loss": 0.0068, + "reward": 1.3336937427520752, + "reward_std": 0.05201172083616257, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.33369380235671997, + "rewards/pad": 0.0, + "step": 354 + }, + { + "completion_length": 96.609375, + "epoch": 0.11312938177182918, + "grad_norm": 57.56117248535156, + "kl": 0.0966796875, + "learning_rate": 8.868706182281707e-07, + "loss": 0.0039, + "reward": 1.5197579860687256, + "reward_std": 0.040330007672309875, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5197579860687256, + "rewards/pad": 0.0, + "step": 355 + }, + { + "completion_length": 70.921875, + "epoch": 0.11344805608667942, + "grad_norm": 53.66902160644531, + "kl": 0.1826171875, + "learning_rate": 8.865519439133205e-07, + "loss": 0.0073, + "reward": 1.4107165336608887, + "reward_std": 0.20225141942501068, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3482164740562439, + "rewards/pad": 0.0625, + "step": 356 + }, + { + "completion_length": 124.359375, + "epoch": 0.11376673040152964, + "grad_norm": 23.252378463745117, + "kl": 0.0732421875, + "learning_rate": 8.862332695984703e-07, + "loss": 0.0029, + "reward": 1.4586352109909058, + "reward_std": 0.09969377517700195, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.474260151386261, + "step": 357 + }, + { + "completion_length": 177.578125, + "epoch": 0.11408540471637986, + "grad_norm": 25.044593811035156, + "kl": 0.05126953125, + "learning_rate": 8.859145952836202e-07, + "loss": 0.0021, + "reward": 1.4991048574447632, + "reward_std": 0.023005153983831406, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.37410491704940796, + "step": 358 + }, + { + "completion_length": 94.765625, + "epoch": 0.11440407903123008, + "grad_norm": 24.907939910888672, + "kl": 0.123046875, + "learning_rate": 8.855959209687699e-07, + "loss": 0.0049, + "reward": 1.4080394506454468, + "reward_std": 0.03388051688671112, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.40803948044776917, + "rewards/pad": 0.0, + "step": 359 + }, + { + "completion_length": 72.53125, + "epoch": 0.1147227533460803, + "grad_norm": 41.2052116394043, + "kl": 0.126953125, + "learning_rate": 8.852772466539197e-07, + "loss": 0.0051, + "reward": 1.5547771453857422, + "reward_std": 0.07238291949033737, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4297771453857422, + "rewards/pad": 0.125, + "step": 360 + }, + { + "completion_length": 98.53125, + "epoch": 0.11504142766093053, + "grad_norm": 47.04529571533203, + "kl": 0.11279296875, + "learning_rate": 8.849585723390695e-07, + "loss": 0.0045, + "reward": 1.6135101318359375, + "reward_std": 0.07768786698579788, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4885101318359375, + "step": 361 + }, + { + "completion_length": 125.328125, + "epoch": 0.11536010197578075, + "grad_norm": 124.77279663085938, + "kl": 0.0673828125, + "learning_rate": 8.846398980242193e-07, + "loss": 0.0027, + "reward": 1.5826597213745117, + "reward_std": 0.09398077428340912, + "rewards/answer_reward": 0.15625, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.42640984058380127, + "step": 362 + }, + { + "completion_length": 22.109375, + "epoch": 0.11567877629063097, + "grad_norm": 55.29496383666992, + "kl": 0.11767578125, + "learning_rate": 8.84321223709369e-07, + "loss": 0.0047, + "reward": 1.9375572204589844, + "reward_std": 0.19804233312606812, + "rewards/answer_reward": 0.515625, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.421932190656662, + "step": 363 + }, + { + "completion_length": 97.78125, + "epoch": 0.1159974506054812, + "grad_norm": 23.788480758666992, + "kl": 0.080078125, + "learning_rate": 8.840025493945188e-07, + "loss": 0.0032, + "reward": 1.5006022453308105, + "reward_std": 0.04466657713055611, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5006022453308105, + "rewards/pad": 0.0, + "step": 364 + }, + { + "completion_length": 146.4375, + "epoch": 0.11631612492033142, + "grad_norm": 20.29625129699707, + "kl": 0.0888671875, + "learning_rate": 8.836838750796686e-07, + "loss": 0.0036, + "reward": 1.5072904825210571, + "reward_std": 0.04846058040857315, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.38229045271873474, + "step": 365 + }, + { + "completion_length": 70.609375, + "epoch": 0.11663479923518165, + "grad_norm": 64.4805679321289, + "kl": 0.1962890625, + "learning_rate": 8.833652007648184e-07, + "loss": 0.0078, + "reward": 1.693876028060913, + "reward_std": 0.14755964279174805, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5688760876655579, + "rewards/pad": 0.125, + "step": 366 + }, + { + "completion_length": 69.875, + "epoch": 0.11695347355003187, + "grad_norm": 35.21623229980469, + "kl": 0.1298828125, + "learning_rate": 8.830465264499681e-07, + "loss": 0.0052, + "reward": 1.5406925678253174, + "reward_std": 0.15816015005111694, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5094425678253174, + "rewards/pad": 0.03125, + "step": 367 + }, + { + "completion_length": 98.671875, + "epoch": 0.11727214786488209, + "grad_norm": 257.2796936035156, + "kl": 0.08935546875, + "learning_rate": 8.827278521351179e-07, + "loss": 0.0036, + "reward": 1.5796436071395874, + "reward_std": 0.14178229868412018, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5327686667442322, + "rewards/pad": 0.046875, + "step": 368 + }, + { + "completion_length": 121.796875, + "epoch": 0.11759082217973231, + "grad_norm": 48.446983337402344, + "kl": 0.083984375, + "learning_rate": 8.824091778202677e-07, + "loss": 0.0033, + "reward": 1.512686014175415, + "reward_std": 0.06343154609203339, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.38768601417541504, + "step": 369 + }, + { + "completion_length": 71.015625, + "epoch": 0.11790949649458253, + "grad_norm": 26.253490447998047, + "kl": 0.125, + "learning_rate": 8.820905035054175e-07, + "loss": 0.005, + "reward": 1.541078805923462, + "reward_std": 0.09650516510009766, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5410787463188171, + "rewards/pad": 0.0, + "step": 370 + }, + { + "completion_length": 47.359375, + "epoch": 0.11822817080943276, + "grad_norm": 41.3243408203125, + "kl": 0.1650390625, + "learning_rate": 8.817718291905672e-07, + "loss": 0.0066, + "reward": 1.6651489734649658, + "reward_std": 0.1509116291999817, + "rewards/answer_reward": 0.234375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.43077388405799866, + "step": 371 + }, + { + "completion_length": 70.703125, + "epoch": 0.11854684512428298, + "grad_norm": 55.84326934814453, + "kl": 0.1875, + "learning_rate": 8.81453154875717e-07, + "loss": 0.0075, + "reward": 1.609081745147705, + "reward_std": 0.12055587768554688, + "rewards/pad": 0.109375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.49970677495002747, + "step": 372 + }, + { + "completion_length": 45.640625, + "epoch": 0.1188655194391332, + "grad_norm": 151.74977111816406, + "kl": 0.1435546875, + "learning_rate": 8.811344805608667e-07, + "loss": 0.0057, + "reward": 1.4652079343795776, + "reward_std": 0.09197273850440979, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4652078449726105, + "step": 373 + }, + { + "completion_length": 44.625, + "epoch": 0.11918419375398343, + "grad_norm": 45.7737922668457, + "kl": 0.1513671875, + "learning_rate": 8.808158062460164e-07, + "loss": 0.0061, + "reward": 1.5594871044158936, + "reward_std": 0.15570798516273499, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5594871640205383, + "rewards/pad": 0.0, + "step": 374 + }, + { + "completion_length": 70.296875, + "epoch": 0.11950286806883365, + "grad_norm": 15.35860824584961, + "kl": 0.1669921875, + "learning_rate": 8.804971319311662e-07, + "loss": 0.0067, + "reward": 1.7419081926345825, + "reward_std": 0.06085880100727081, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6169082522392273, + "rewards/pad": 0.125, + "step": 375 + }, + { + "completion_length": 73.140625, + "epoch": 0.11982154238368388, + "grad_norm": 43.59029006958008, + "kl": 0.11181640625, + "learning_rate": 8.80178457616316e-07, + "loss": 0.0045, + "reward": 1.6957318782806396, + "reward_std": 0.09901833534240723, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4457317590713501, + "rewards/pad": 0.25, + "step": 376 + }, + { + "completion_length": 151.546875, + "epoch": 0.1201402166985341, + "grad_norm": 44.84366989135742, + "kl": 0.05126953125, + "learning_rate": 8.798597833014659e-07, + "loss": 0.0021, + "reward": 1.5476773977279663, + "reward_std": 0.061405330896377563, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4226773977279663, + "step": 377 + }, + { + "completion_length": 74.0625, + "epoch": 0.12045889101338432, + "grad_norm": 33.01712417602539, + "kl": 0.09521484375, + "learning_rate": 8.795411089866156e-07, + "loss": 0.0038, + "reward": 1.4716873168945312, + "reward_std": 0.13285109400749207, + "rewards/pad": 0.09375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3779374063014984, + "step": 378 + }, + { + "completion_length": 46.515625, + "epoch": 0.12077756532823454, + "grad_norm": 36.0592041015625, + "kl": 0.1064453125, + "learning_rate": 8.792224346717654e-07, + "loss": 0.0042, + "reward": 1.7025115489959717, + "reward_std": 0.18063417077064514, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5306365489959717, + "rewards/pad": 0.171875, + "step": 379 + }, + { + "completion_length": 97.40625, + "epoch": 0.12109623964308477, + "grad_norm": 63.695777893066406, + "kl": 0.0859375, + "learning_rate": 8.789037603569152e-07, + "loss": 0.0034, + "reward": 1.573782205581665, + "reward_std": 0.11549675464630127, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.44878220558166504, + "step": 380 + }, + { + "completion_length": 97.625, + "epoch": 0.121414913957935, + "grad_norm": 45.81637954711914, + "kl": 0.326171875, + "learning_rate": 8.78585086042065e-07, + "loss": 0.013, + "reward": 1.4872021675109863, + "reward_std": 0.10220817476511002, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4872022271156311, + "rewards/pad": 0.0, + "step": 381 + }, + { + "completion_length": 97.0, + "epoch": 0.12173358827278521, + "grad_norm": 79.03022766113281, + "kl": 0.08935546875, + "learning_rate": 8.782664117272147e-07, + "loss": 0.0036, + "reward": 1.592256784439087, + "reward_std": 0.09763297438621521, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.46725672483444214, + "step": 382 + }, + { + "completion_length": 98.640625, + "epoch": 0.12205226258763544, + "grad_norm": 81.4788818359375, + "kl": 0.0947265625, + "learning_rate": 8.779477374123645e-07, + "loss": 0.0038, + "reward": 1.4392238855361938, + "reward_std": 0.10192390531301498, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.31422388553619385, + "rewards/pad": 0.125, + "step": 383 + }, + { + "completion_length": 123.609375, + "epoch": 0.12237093690248566, + "grad_norm": 61.001285552978516, + "kl": 0.08740234375, + "learning_rate": 8.776290630975143e-07, + "loss": 0.0035, + "reward": 1.4689092636108398, + "reward_std": 0.057621996849775314, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3439093828201294, + "rewards/pad": 0.125, + "step": 384 + }, + { + "completion_length": 98.40625, + "epoch": 0.12268961121733589, + "grad_norm": 28.01368522644043, + "kl": 0.12890625, + "learning_rate": 8.773103887826641e-07, + "loss": 0.0051, + "reward": 1.4256947040557861, + "reward_std": 0.08606939017772675, + "rewards/answer_reward": 0.0, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.42569467425346375, + "step": 385 + }, + { + "completion_length": 71.609375, + "epoch": 0.12300828553218611, + "grad_norm": 29.05324363708496, + "kl": 0.1259765625, + "learning_rate": 8.769917144678138e-07, + "loss": 0.005, + "reward": 1.6970813274383545, + "reward_std": 0.06371637433767319, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5720812082290649, + "rewards/pad": 0.125, + "step": 386 + }, + { + "completion_length": 46.125, + "epoch": 0.12332695984703633, + "grad_norm": 37.76282501220703, + "kl": 0.1279296875, + "learning_rate": 8.766730401529636e-07, + "loss": 0.0051, + "reward": 1.5817197561264038, + "reward_std": 0.12376982718706131, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4567198157310486, + "rewards/pad": 0.125, + "step": 387 + }, + { + "completion_length": 74.953125, + "epoch": 0.12364563416188655, + "grad_norm": 29.852304458618164, + "kl": 0.09228515625, + "learning_rate": 8.763543658381134e-07, + "loss": 0.0037, + "reward": 1.7793254852294922, + "reward_std": 0.10431215912103653, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4355755150318146, + "rewards/pad": 0.34375, + "step": 388 + }, + { + "completion_length": 123.40625, + "epoch": 0.12396430847673677, + "grad_norm": 95.40129852294922, + "kl": 0.06103515625, + "learning_rate": 8.760356915232632e-07, + "loss": 0.0024, + "reward": 1.5170669555664062, + "reward_std": 0.19634395837783813, + "rewards/pad": 0.171875, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.3608168363571167, + "step": 389 + }, + { + "completion_length": 151.34375, + "epoch": 0.124282982791587, + "grad_norm": 14.677362442016602, + "kl": 0.087890625, + "learning_rate": 8.757170172084129e-07, + "loss": 0.0035, + "reward": 1.4010517597198486, + "reward_std": 0.02272605523467064, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.2760518193244934, + "step": 390 + }, + { + "completion_length": 70.28125, + "epoch": 0.12460165710643722, + "grad_norm": 32.54584884643555, + "kl": 0.150390625, + "learning_rate": 8.753983428935627e-07, + "loss": 0.006, + "reward": 1.6120429039001465, + "reward_std": 0.08093911409378052, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.48704293370246887, + "step": 391 + }, + { + "completion_length": 148.484375, + "epoch": 0.12492033142128744, + "grad_norm": 22.513427734375, + "kl": 0.06298828125, + "learning_rate": 8.750796685787125e-07, + "loss": 0.0025, + "reward": 1.3614397048950195, + "reward_std": 0.05563504248857498, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3614397644996643, + "step": 392 + }, + { + "completion_length": 124.734375, + "epoch": 0.12523900573613767, + "grad_norm": 27.51471519470215, + "kl": 0.0791015625, + "learning_rate": 8.747609942638623e-07, + "loss": 0.0032, + "reward": 1.738678216934204, + "reward_std": 0.1270914524793625, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 0.984375, + "rewards/iou_glue_reward": 0.5043032169342041, + "step": 393 + }, + { + "completion_length": 124.046875, + "epoch": 0.12555768005098789, + "grad_norm": 78.34660339355469, + "kl": 0.06201171875, + "learning_rate": 8.74442319949012e-07, + "loss": 0.0025, + "reward": 1.4761624336242676, + "reward_std": 0.08410072326660156, + "rewards/answer_reward": 0.109375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.3667874038219452, + "step": 394 + }, + { + "completion_length": 177.921875, + "epoch": 0.1258763543658381, + "grad_norm": 38.84751892089844, + "kl": 0.1650390625, + "learning_rate": 8.741236456341619e-07, + "loss": 0.0066, + "reward": 1.2327470779418945, + "reward_std": 0.019952615723013878, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.23274700343608856, + "step": 395 + }, + { + "completion_length": 46.65625, + "epoch": 0.12619502868068833, + "grad_norm": 29.15982437133789, + "kl": 0.1494140625, + "learning_rate": 8.738049713193117e-07, + "loss": 0.0059, + "reward": 1.751692533493042, + "reward_std": 0.07417924702167511, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5016926527023315, + "step": 396 + }, + { + "completion_length": 45.5, + "epoch": 0.12651370299553855, + "grad_norm": 53.9080696105957, + "kl": 0.0947265625, + "learning_rate": 8.734862970044615e-07, + "loss": 0.0038, + "reward": 1.654681921005249, + "reward_std": 0.14040318131446838, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.48280686140060425, + "rewards/pad": 0.171875, + "step": 397 + }, + { + "completion_length": 124.953125, + "epoch": 0.1268323773103888, + "grad_norm": 11.394405364990234, + "kl": 0.064453125, + "learning_rate": 8.731676226896112e-07, + "loss": 0.0026, + "reward": 1.7598766088485718, + "reward_std": 0.08118850737810135, + "rewards/answer_reward": 0.359375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4005016088485718, + "step": 398 + }, + { + "completion_length": 121.28125, + "epoch": 0.12715105162523901, + "grad_norm": 107.51454162597656, + "kl": 0.07958984375, + "learning_rate": 8.72848948374761e-07, + "loss": 0.0032, + "reward": 1.3193341493606567, + "reward_std": 0.06677798926830292, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3193341791629791, + "step": 399 + }, + { + "completion_length": 19.828125, + "epoch": 0.12746972594008923, + "grad_norm": 31.021102905273438, + "kl": 0.2021484375, + "learning_rate": 8.725302740599108e-07, + "loss": 0.0081, + "reward": 1.7511768341064453, + "reward_std": 0.14500945806503296, + "rewards/answer_reward": 0.234375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5168018937110901, + "step": 400 + }, + { + "completion_length": 148.140625, + "epoch": 0.12778840025493945, + "grad_norm": 16.07240104675293, + "kl": 0.09375, + "learning_rate": 8.722115997450606e-07, + "loss": 0.0038, + "reward": 1.3646976947784424, + "reward_std": 0.03278578072786331, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.36469757556915283, + "step": 401 + }, + { + "completion_length": 45.4375, + "epoch": 0.12810707456978968, + "grad_norm": 40.273704528808594, + "kl": 0.15234375, + "learning_rate": 8.718929254302103e-07, + "loss": 0.0061, + "reward": 1.6111754179000854, + "reward_std": 0.09451623260974884, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4861753582954407, + "step": 402 + }, + { + "completion_length": 123.953125, + "epoch": 0.1284257488846399, + "grad_norm": 25.872350692749023, + "kl": 0.07275390625, + "learning_rate": 8.715742511153601e-07, + "loss": 0.0029, + "reward": 1.568497657775879, + "reward_std": 0.07783734798431396, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.44349759817123413, + "rewards/pad": 0.125, + "step": 403 + }, + { + "completion_length": 68.53125, + "epoch": 0.12874442319949012, + "grad_norm": 25.39286231994629, + "kl": 0.1826171875, + "learning_rate": 8.712555768005099e-07, + "loss": 0.0073, + "reward": 1.4752670526504517, + "reward_std": 0.15198847651481628, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.49089211225509644, + "rewards/pad": 0.0, + "step": 404 + }, + { + "completion_length": 71.8125, + "epoch": 0.12906309751434034, + "grad_norm": 177.87518310546875, + "kl": 0.1103515625, + "learning_rate": 8.709369024856596e-07, + "loss": 0.0044, + "reward": 1.634433627128601, + "reward_std": 0.10617288947105408, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5250586271286011, + "rewards/pad": 0.109375, + "step": 405 + }, + { + "completion_length": 100.03125, + "epoch": 0.12938177182919056, + "grad_norm": 33.5173454284668, + "kl": 0.158203125, + "learning_rate": 8.706182281708094e-07, + "loss": 0.0063, + "reward": 1.6183446645736694, + "reward_std": 0.12125556170940399, + "rewards/pad": 0.140625, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.47771957516670227, + "step": 406 + }, + { + "completion_length": 73.484375, + "epoch": 0.1297004461440408, + "grad_norm": 101.80476379394531, + "kl": 0.162109375, + "learning_rate": 8.702995538559592e-07, + "loss": 0.0065, + "reward": 1.7018132209777832, + "reward_std": 0.13162967562675476, + "rewards/pad": 0.21875, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4830631613731384, + "step": 407 + }, + { + "completion_length": 95.15625, + "epoch": 0.13001912045889102, + "grad_norm": 36.46905517578125, + "kl": 0.125, + "learning_rate": 8.69980879541109e-07, + "loss": 0.005, + "reward": 1.3658957481384277, + "reward_std": 0.06615029275417328, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.36589574813842773, + "step": 408 + }, + { + "completion_length": 174.375, + "epoch": 0.13033779477374124, + "grad_norm": 30.588741302490234, + "kl": 0.58203125, + "learning_rate": 8.696622052262587e-07, + "loss": 0.0233, + "reward": 1.4599390029907227, + "reward_std": 0.09191317856311798, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.45993906259536743, + "step": 409 + }, + { + "completion_length": 72.21875, + "epoch": 0.13065646908859146, + "grad_norm": 28.82553482055664, + "kl": 0.12890625, + "learning_rate": 8.693435309114085e-07, + "loss": 0.0052, + "reward": 1.7069793939590454, + "reward_std": 0.10922467708587646, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4569793939590454, + "step": 410 + }, + { + "completion_length": 20.109375, + "epoch": 0.13097514340344169, + "grad_norm": 31.949430465698242, + "kl": 0.15234375, + "learning_rate": 8.690248565965583e-07, + "loss": 0.0061, + "reward": 1.583141565322876, + "reward_std": 0.07516056299209595, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.458141565322876, + "rewards/pad": 0.125, + "step": 411 + }, + { + "completion_length": 96.875, + "epoch": 0.1312938177182919, + "grad_norm": 21.5670223236084, + "kl": 0.09375, + "learning_rate": 8.68706182281708e-07, + "loss": 0.0037, + "reward": 1.437939167022705, + "reward_std": 0.10327853262424469, + "rewards/answer_reward": 0.0, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4379391074180603, + "step": 412 + }, + { + "completion_length": 100.40625, + "epoch": 0.13161249203314213, + "grad_norm": 55.93552017211914, + "kl": 0.0859375, + "learning_rate": 8.683875079668577e-07, + "loss": 0.0034, + "reward": 1.4741472005844116, + "reward_std": 0.05575428530573845, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.34914714097976685, + "rewards/pad": 0.125, + "step": 413 + }, + { + "completion_length": 98.203125, + "epoch": 0.13193116634799235, + "grad_norm": 121.39154052734375, + "kl": 0.1279296875, + "learning_rate": 8.680688336520075e-07, + "loss": 0.0051, + "reward": 1.4627301692962646, + "reward_std": 0.09106460213661194, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.33773019909858704, + "step": 414 + }, + { + "completion_length": 98.90625, + "epoch": 0.13224984066284257, + "grad_norm": 37.05792999267578, + "kl": 0.0830078125, + "learning_rate": 8.677501593371574e-07, + "loss": 0.0033, + "reward": 1.5707786083221436, + "reward_std": 0.11219216883182526, + "rewards/answer_reward": 0.0625, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5082787275314331, + "step": 415 + }, + { + "completion_length": 46.234375, + "epoch": 0.1325685149776928, + "grad_norm": 26.515222549438477, + "kl": 0.146484375, + "learning_rate": 8.674314850223072e-07, + "loss": 0.0059, + "reward": 1.4971379041671753, + "reward_std": 0.09204782545566559, + "rewards/answer_reward": 0.0, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4971378445625305, + "step": 416 + }, + { + "completion_length": 97.828125, + "epoch": 0.13288718929254303, + "grad_norm": 48.868499755859375, + "kl": 0.07080078125, + "learning_rate": 8.671128107074569e-07, + "loss": 0.0028, + "reward": 1.7335339784622192, + "reward_std": 0.12875020503997803, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.499159038066864, + "rewards/pad": 0.234375, + "step": 417 + }, + { + "completion_length": 70.671875, + "epoch": 0.13320586360739325, + "grad_norm": 33.53970718383789, + "kl": 0.1474609375, + "learning_rate": 8.667941363926067e-07, + "loss": 0.0059, + "reward": 1.6477628946304321, + "reward_std": 0.11864393204450607, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5383878946304321, + "rewards/pad": 0.109375, + "step": 418 + }, + { + "completion_length": 94.390625, + "epoch": 0.13352453792224347, + "grad_norm": 44.05862045288086, + "kl": 0.11328125, + "learning_rate": 8.664754620777565e-07, + "loss": 0.0045, + "reward": 1.4214290380477905, + "reward_std": 0.08193930238485336, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.42142897844314575, + "step": 419 + }, + { + "completion_length": 70.859375, + "epoch": 0.1338432122370937, + "grad_norm": 16.737979888916016, + "kl": 0.12451171875, + "learning_rate": 8.661567877629063e-07, + "loss": 0.005, + "reward": 1.6469398736953735, + "reward_std": 0.07638402283191681, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6469398736953735, + "rewards/pad": 0.0, + "step": 420 + }, + { + "completion_length": 120.421875, + "epoch": 0.13416188655194392, + "grad_norm": 26.583585739135742, + "kl": 0.2138671875, + "learning_rate": 8.65838113448056e-07, + "loss": 0.0086, + "reward": 1.4169089794158936, + "reward_std": 0.07294677197933197, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.41690900921821594, + "rewards/pad": 0.0, + "step": 421 + }, + { + "completion_length": 97.40625, + "epoch": 0.13448056086679414, + "grad_norm": 21.15988540649414, + "kl": 0.068359375, + "learning_rate": 8.655194391332058e-07, + "loss": 0.0027, + "reward": 1.4524857997894287, + "reward_std": 0.054185282438993454, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4524857997894287, + "rewards/pad": 0.0, + "step": 422 + }, + { + "completion_length": 46.53125, + "epoch": 0.13479923518164436, + "grad_norm": 90.01542663574219, + "kl": 0.123046875, + "learning_rate": 8.652007648183556e-07, + "loss": 0.0049, + "reward": 1.6122829914093018, + "reward_std": 0.11555016785860062, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.48728299140930176, + "rewards/pad": 0.125, + "step": 423 + }, + { + "completion_length": 94.4375, + "epoch": 0.13511790949649458, + "grad_norm": 30.43792724609375, + "kl": 0.0908203125, + "learning_rate": 8.648820905035054e-07, + "loss": 0.0036, + "reward": 1.5253713130950928, + "reward_std": 0.06750549376010895, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5253711938858032, + "rewards/pad": 0.0, + "step": 424 + }, + { + "completion_length": 98.25, + "epoch": 0.1354365838113448, + "grad_norm": 50.584659576416016, + "kl": 0.125, + "learning_rate": 8.645634161886551e-07, + "loss": 0.005, + "reward": 1.4811164140701294, + "reward_std": 0.04695218801498413, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4811164438724518, + "step": 425 + }, + { + "completion_length": 174.84375, + "epoch": 0.13575525812619502, + "grad_norm": 26.733871459960938, + "kl": 0.053466796875, + "learning_rate": 8.642447418738049e-07, + "loss": 0.0021, + "reward": 1.3371973037719727, + "reward_std": 0.024800507351756096, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3371972441673279, + "rewards/pad": 0.0, + "step": 426 + }, + { + "completion_length": 98.34375, + "epoch": 0.13607393244104526, + "grad_norm": 36.893760681152344, + "kl": 0.11962890625, + "learning_rate": 8.639260675589547e-07, + "loss": 0.0048, + "reward": 1.6948579549789429, + "reward_std": 0.10014330595731735, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5698579549789429, + "rewards/pad": 0.125, + "step": 427 + }, + { + "completion_length": 122.75, + "epoch": 0.13639260675589548, + "grad_norm": 152.30303955078125, + "kl": 0.12158203125, + "learning_rate": 8.636073932441045e-07, + "loss": 0.0049, + "reward": 1.5262490510940552, + "reward_std": 0.07107570022344589, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4012489914894104, + "step": 428 + }, + { + "completion_length": 96.6875, + "epoch": 0.1367112810707457, + "grad_norm": 40.88010787963867, + "kl": 0.12890625, + "learning_rate": 8.632887189292542e-07, + "loss": 0.0051, + "reward": 1.2539267539978027, + "reward_std": 0.09933836758136749, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.2539267838001251, + "rewards/pad": 0.0, + "step": 429 + }, + { + "completion_length": 72.09375, + "epoch": 0.13702995538559593, + "grad_norm": 58.52267074584961, + "kl": 0.162109375, + "learning_rate": 8.62970044614404e-07, + "loss": 0.0065, + "reward": 1.5780236721038818, + "reward_std": 0.05143863707780838, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.32802361249923706, + "step": 430 + }, + { + "completion_length": 45.765625, + "epoch": 0.13734862970044615, + "grad_norm": 31.6570987701416, + "kl": 0.1318359375, + "learning_rate": 8.626513702995538e-07, + "loss": 0.0053, + "reward": 1.4814674854278564, + "reward_std": 0.05928094685077667, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.35646751523017883, + "rewards/pad": 0.125, + "step": 431 + }, + { + "completion_length": 153.484375, + "epoch": 0.13766730401529637, + "grad_norm": 109.1246337890625, + "kl": 0.09521484375, + "learning_rate": 8.623326959847035e-07, + "loss": 0.0038, + "reward": 1.4952130317687988, + "reward_std": 0.0844242125749588, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.49521297216415405, + "rewards/pad": 0.0, + "step": 432 + }, + { + "completion_length": 47.296875, + "epoch": 0.13798597833014659, + "grad_norm": 52.604557037353516, + "kl": 0.14453125, + "learning_rate": 8.620140216698534e-07, + "loss": 0.0058, + "reward": 1.4703024625778198, + "reward_std": 0.16446053981781006, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3453024625778198, + "rewards/pad": 0.125, + "step": 433 + }, + { + "completion_length": 175.828125, + "epoch": 0.1383046526449968, + "grad_norm": 21.99553108215332, + "kl": 0.0400390625, + "learning_rate": 8.616953473550032e-07, + "loss": 0.0016, + "reward": 1.3667819499969482, + "reward_std": 0.03961355984210968, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.36678197979927063, + "rewards/pad": 0.0, + "step": 434 + }, + { + "completion_length": 97.375, + "epoch": 0.13862332695984703, + "grad_norm": 69.48381042480469, + "kl": 0.11376953125, + "learning_rate": 8.61376673040153e-07, + "loss": 0.0045, + "reward": 1.5398564338684082, + "reward_std": 0.08691239356994629, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.41485652327537537, + "step": 435 + }, + { + "completion_length": 120.765625, + "epoch": 0.13894200127469725, + "grad_norm": 52.82097625732422, + "kl": 0.15625, + "learning_rate": 8.610579987253027e-07, + "loss": 0.0062, + "reward": 1.4473316669464111, + "reward_std": 0.09470543265342712, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.44733160734176636, + "step": 436 + }, + { + "completion_length": 124.859375, + "epoch": 0.1392606755895475, + "grad_norm": 14.809861183166504, + "kl": 0.1123046875, + "learning_rate": 8.607393244104525e-07, + "loss": 0.0045, + "reward": 1.487265706062317, + "reward_std": 0.06385494023561478, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.2372657209634781, + "step": 437 + }, + { + "completion_length": 43.8125, + "epoch": 0.13957934990439771, + "grad_norm": 24.490432739257812, + "kl": 0.14453125, + "learning_rate": 8.604206500956023e-07, + "loss": 0.0058, + "reward": 1.5618922710418701, + "reward_std": 0.08179674297571182, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5618922710418701, + "rewards/pad": 0.0, + "step": 438 + }, + { + "completion_length": 45.671875, + "epoch": 0.13989802421924794, + "grad_norm": 37.8846549987793, + "kl": 0.13671875, + "learning_rate": 8.601019757807521e-07, + "loss": 0.0055, + "reward": 1.68092942237854, + "reward_std": 0.16623428463935852, + "rewards/answer_reward": 0.234375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4465544819831848, + "step": 439 + }, + { + "completion_length": 94.625, + "epoch": 0.14021669853409816, + "grad_norm": 27.170732498168945, + "kl": 0.09130859375, + "learning_rate": 8.597833014659018e-07, + "loss": 0.0037, + "reward": 1.446689248085022, + "reward_std": 0.089576356112957, + "rewards/answer_reward": 0.0, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.446689248085022, + "step": 440 + }, + { + "completion_length": 69.03125, + "epoch": 0.14053537284894838, + "grad_norm": 51.69449234008789, + "kl": 0.1025390625, + "learning_rate": 8.594646271510516e-07, + "loss": 0.0041, + "reward": 1.4989007711410522, + "reward_std": 0.08558598160743713, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.37390074133872986, + "rewards/pad": 0.125, + "step": 441 + }, + { + "completion_length": 122.75, + "epoch": 0.1408540471637986, + "grad_norm": 20.52138900756836, + "kl": 0.1396484375, + "learning_rate": 8.591459528362014e-07, + "loss": 0.0056, + "reward": 1.5021915435791016, + "reward_std": 0.08817140758037567, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5021916031837463, + "rewards/pad": 0.0, + "step": 442 + }, + { + "completion_length": 69.15625, + "epoch": 0.14117272147864882, + "grad_norm": 69.19669342041016, + "kl": 0.2158203125, + "learning_rate": 8.588272785213512e-07, + "loss": 0.0086, + "reward": 1.5405244827270508, + "reward_std": 0.07173141837120056, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4155244529247284, + "rewards/pad": 0.125, + "step": 443 + }, + { + "completion_length": 122.1875, + "epoch": 0.14149139579349904, + "grad_norm": 26.565298080444336, + "kl": 0.078125, + "learning_rate": 8.585086042065009e-07, + "loss": 0.0031, + "reward": 1.5234211683273315, + "reward_std": 0.044328056275844574, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5234212279319763, + "step": 444 + }, + { + "completion_length": 72.453125, + "epoch": 0.14181007010834926, + "grad_norm": 26.43044090270996, + "kl": 0.103515625, + "learning_rate": 8.581899298916507e-07, + "loss": 0.0041, + "reward": 1.720982313156128, + "reward_std": 0.05353473871946335, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5959822535514832, + "rewards/pad": 0.125, + "step": 445 + }, + { + "completion_length": 123.890625, + "epoch": 0.14212874442319948, + "grad_norm": 25.059282302856445, + "kl": 0.09130859375, + "learning_rate": 8.578712555768005e-07, + "loss": 0.0037, + "reward": 1.4826295375823975, + "reward_std": 0.06169476360082626, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4826295077800751, + "step": 446 + }, + { + "completion_length": 151.625, + "epoch": 0.14244741873804972, + "grad_norm": 17.98177719116211, + "kl": 0.068359375, + "learning_rate": 8.575525812619503e-07, + "loss": 0.0027, + "reward": 1.389360785484314, + "reward_std": 0.04530121758580208, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3893607258796692, + "step": 447 + }, + { + "completion_length": 71.0, + "epoch": 0.14276609305289995, + "grad_norm": 25.77447509765625, + "kl": 0.1806640625, + "learning_rate": 8.572339069471e-07, + "loss": 0.0072, + "reward": 1.6751841306686401, + "reward_std": 0.057732950896024704, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6751840710639954, + "rewards/pad": 0.0, + "step": 448 + }, + { + "completion_length": 151.703125, + "epoch": 0.14308476736775017, + "grad_norm": 17.309431076049805, + "kl": 0.11669921875, + "learning_rate": 8.569152326322498e-07, + "loss": 0.0047, + "reward": 1.6690850257873535, + "reward_std": 0.058293867856264114, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5440850257873535, + "step": 449 + }, + { + "completion_length": 67.96875, + "epoch": 0.14340344168260039, + "grad_norm": 43.26914978027344, + "kl": 0.13671875, + "learning_rate": 8.565965583173996e-07, + "loss": 0.0055, + "reward": 1.5374586582183838, + "reward_std": 0.07611675560474396, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5374586582183838, + "rewards/pad": 0.0, + "step": 450 + }, + { + "completion_length": 49.984375, + "epoch": 0.1437221159974506, + "grad_norm": 38.74765396118164, + "kl": 0.0830078125, + "learning_rate": 8.562778840025495e-07, + "loss": 0.0033, + "reward": 1.665922999382019, + "reward_std": 0.1814975142478943, + "rewards/answer_reward": 0.28125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.38467299938201904, + "step": 451 + }, + { + "completion_length": 72.046875, + "epoch": 0.14404079031230083, + "grad_norm": 183.16456604003906, + "kl": 0.1533203125, + "learning_rate": 8.55959209687699e-07, + "loss": 0.0061, + "reward": 1.5401326417922974, + "reward_std": 0.05094554275274277, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.41513270139694214, + "step": 452 + }, + { + "completion_length": 72.515625, + "epoch": 0.14435946462715105, + "grad_norm": 242.4665069580078, + "kl": 0.11669921875, + "learning_rate": 8.556405353728489e-07, + "loss": 0.0047, + "reward": 1.6754543781280518, + "reward_std": 0.0814475268125534, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4254542887210846, + "rewards/pad": 0.25, + "step": 453 + }, + { + "completion_length": 176.53125, + "epoch": 0.14467813894200127, + "grad_norm": 62.67466354370117, + "kl": 0.0625, + "learning_rate": 8.553218610579987e-07, + "loss": 0.0025, + "reward": 1.4424443244934082, + "reward_std": 0.05421048775315285, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.442444384098053, + "step": 454 + }, + { + "completion_length": 124.0625, + "epoch": 0.1449968132568515, + "grad_norm": 27.845863342285156, + "kl": 0.107421875, + "learning_rate": 8.550031867431485e-07, + "loss": 0.0043, + "reward": 1.4571428298950195, + "reward_std": 0.07115037739276886, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.45714280009269714, + "rewards/pad": 0.0, + "step": 455 + }, + { + "completion_length": 99.859375, + "epoch": 0.14531548757170173, + "grad_norm": 18.424503326416016, + "kl": 0.10791015625, + "learning_rate": 8.546845124282982e-07, + "loss": 0.0043, + "reward": 1.5426874160766602, + "reward_std": 0.04801858961582184, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.41768738627433777, + "rewards/pad": 0.125, + "step": 456 + }, + { + "completion_length": 99.546875, + "epoch": 0.14563416188655195, + "grad_norm": 42.82892608642578, + "kl": 0.2041015625, + "learning_rate": 8.54365838113448e-07, + "loss": 0.0082, + "reward": 1.4242668151855469, + "reward_std": 0.07840073853731155, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4242668151855469, + "step": 457 + }, + { + "completion_length": 47.53125, + "epoch": 0.14595283620140218, + "grad_norm": 83.61685180664062, + "kl": 0.1259765625, + "learning_rate": 8.540471637985978e-07, + "loss": 0.005, + "reward": 1.5576732158660889, + "reward_std": 0.1142008900642395, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.43267327547073364, + "rewards/pad": 0.125, + "step": 458 + }, + { + "completion_length": 72.328125, + "epoch": 0.1462715105162524, + "grad_norm": 36.45909881591797, + "kl": 0.1318359375, + "learning_rate": 8.537284894837476e-07, + "loss": 0.0053, + "reward": 1.6129789352416992, + "reward_std": 0.06982880085706711, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.612978994846344, + "rewards/pad": 0.0, + "step": 459 + }, + { + "completion_length": 74.953125, + "epoch": 0.14659018483110262, + "grad_norm": 43.895931243896484, + "kl": 0.1328125, + "learning_rate": 8.534098151688973e-07, + "loss": 0.0053, + "reward": 1.4296586513519287, + "reward_std": 0.136674702167511, + "rewards/pad": 0.03125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3984086215496063, + "step": 460 + }, + { + "completion_length": 73.40625, + "epoch": 0.14690885914595284, + "grad_norm": 453.3140869140625, + "kl": 0.1181640625, + "learning_rate": 8.530911408540471e-07, + "loss": 0.0047, + "reward": 1.6739246845245361, + "reward_std": 0.11186471581459045, + "rewards/answer_reward": 0.171875, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5020498037338257, + "step": 461 + }, + { + "completion_length": 99.703125, + "epoch": 0.14722753346080306, + "grad_norm": 98.9237060546875, + "kl": 0.09423828125, + "learning_rate": 8.527724665391969e-07, + "loss": 0.0038, + "reward": 1.5939929485321045, + "reward_std": 0.06156596168875694, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4689928889274597, + "step": 462 + }, + { + "completion_length": 177.03125, + "epoch": 0.14754620777565328, + "grad_norm": 12.542116165161133, + "kl": 0.06201171875, + "learning_rate": 8.524537922243466e-07, + "loss": 0.0025, + "reward": 1.4433605670928955, + "reward_std": 0.04455003887414932, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4433605670928955, + "step": 463 + }, + { + "completion_length": 179.28125, + "epoch": 0.1478648820905035, + "grad_norm": 31.899578094482422, + "kl": 0.052734375, + "learning_rate": 8.521351179094964e-07, + "loss": 0.0021, + "reward": 1.3155629634857178, + "reward_std": 0.07355151325464249, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3155629336833954, + "step": 464 + }, + { + "completion_length": 45.609375, + "epoch": 0.14818355640535372, + "grad_norm": 77.76405334472656, + "kl": 0.1689453125, + "learning_rate": 8.518164435946462e-07, + "loss": 0.0068, + "reward": 1.4881693124771118, + "reward_std": 0.09249794483184814, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4881693124771118, + "rewards/pad": 0.0, + "step": 465 + }, + { + "completion_length": 73.953125, + "epoch": 0.14850223072020396, + "grad_norm": 73.53346252441406, + "kl": 0.35546875, + "learning_rate": 8.51497769279796e-07, + "loss": 0.0142, + "reward": 1.422756314277649, + "reward_std": 0.1399630308151245, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.2821313142776489, + "rewards/pad": 0.140625, + "step": 466 + }, + { + "completion_length": 70.421875, + "epoch": 0.14882090503505419, + "grad_norm": 33.81299591064453, + "kl": 0.125, + "learning_rate": 8.511790949649457e-07, + "loss": 0.005, + "reward": 1.705796718597412, + "reward_std": 0.07710770517587662, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5807967782020569, + "rewards/pad": 0.125, + "step": 467 + }, + { + "completion_length": 125.328125, + "epoch": 0.1491395793499044, + "grad_norm": 191.01551818847656, + "kl": 0.09765625, + "learning_rate": 8.508604206500955e-07, + "loss": 0.0039, + "reward": 1.4011149406433105, + "reward_std": 0.10568197071552277, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.4167398512363434, + "rewards/pad": 0.0, + "step": 468 + }, + { + "completion_length": 101.484375, + "epoch": 0.14945825366475463, + "grad_norm": 21.767581939697266, + "kl": 0.291015625, + "learning_rate": 8.505417463352453e-07, + "loss": 0.0116, + "reward": 1.6511127948760986, + "reward_std": 0.2230065017938614, + "rewards/pad": 0.21875, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.44798779487609863, + "step": 469 + }, + { + "completion_length": 77.421875, + "epoch": 0.14977692797960485, + "grad_norm": 54.641815185546875, + "kl": 0.1357421875, + "learning_rate": 8.502230720203951e-07, + "loss": 0.0054, + "reward": 1.6336212158203125, + "reward_std": 0.18447574973106384, + "rewards/answer_reward": 0.234375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.3992462754249573, + "step": 470 + }, + { + "completion_length": 99.40625, + "epoch": 0.15009560229445507, + "grad_norm": 21.85428810119629, + "kl": 0.10888671875, + "learning_rate": 8.499043977055449e-07, + "loss": 0.0044, + "reward": 1.6280136108398438, + "reward_std": 0.13695479929447174, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.5186384916305542, + "rewards/pad": 0.125, + "step": 471 + }, + { + "completion_length": 44.84375, + "epoch": 0.1504142766093053, + "grad_norm": 31.97594451904297, + "kl": 0.1357421875, + "learning_rate": 8.495857233906947e-07, + "loss": 0.0054, + "reward": 1.5573021173477173, + "reward_std": 0.051747020334005356, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5573021769523621, + "rewards/pad": 0.0, + "step": 472 + }, + { + "completion_length": 128.71875, + "epoch": 0.1507329509241555, + "grad_norm": 43.0109977722168, + "kl": 0.1357421875, + "learning_rate": 8.492670490758445e-07, + "loss": 0.0054, + "reward": 1.5580388307571411, + "reward_std": 0.11918258666992188, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4174138903617859, + "rewards/pad": 0.140625, + "step": 473 + }, + { + "completion_length": 99.21875, + "epoch": 0.15105162523900573, + "grad_norm": 43.45140838623047, + "kl": 0.091796875, + "learning_rate": 8.489483747609943e-07, + "loss": 0.0037, + "reward": 1.6057767868041992, + "reward_std": 0.08008240908384323, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.6057767868041992, + "step": 474 + }, + { + "completion_length": 124.359375, + "epoch": 0.15137029955385595, + "grad_norm": 113.9756088256836, + "kl": 0.08642578125, + "learning_rate": 8.48629700446144e-07, + "loss": 0.0035, + "reward": 1.5617237091064453, + "reward_std": 0.061002567410469055, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5617236495018005, + "step": 475 + }, + { + "completion_length": 102.765625, + "epoch": 0.1516889738687062, + "grad_norm": 57.76559829711914, + "kl": 0.1328125, + "learning_rate": 8.483110261312938e-07, + "loss": 0.0053, + "reward": 1.4934542179107666, + "reward_std": 0.07229776680469513, + "rewards/pad": 0.109375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.38407933712005615, + "step": 476 + }, + { + "completion_length": 72.296875, + "epoch": 0.15200764818355642, + "grad_norm": 21.623546600341797, + "kl": 0.10595703125, + "learning_rate": 8.479923518164436e-07, + "loss": 0.0042, + "reward": 1.6081228256225586, + "reward_std": 0.08545694500207901, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4831227958202362, + "step": 477 + }, + { + "completion_length": 101.859375, + "epoch": 0.15232632249840664, + "grad_norm": 116.93062591552734, + "kl": 0.11279296875, + "learning_rate": 8.476736775015934e-07, + "loss": 0.0045, + "reward": 1.6542677879333496, + "reward_std": 0.14931340515613556, + "rewards/answer_reward": 0.171875, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.482392817735672, + "step": 478 + }, + { + "completion_length": 125.34375, + "epoch": 0.15264499681325686, + "grad_norm": 107.00728607177734, + "kl": 0.142578125, + "learning_rate": 8.473550031867431e-07, + "loss": 0.0057, + "reward": 1.472907304763794, + "reward_std": 0.06427513062953949, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4729074239730835, + "step": 479 + }, + { + "completion_length": 102.4375, + "epoch": 0.15296367112810708, + "grad_norm": 18.603273391723633, + "kl": 0.1142578125, + "learning_rate": 8.470363288718929e-07, + "loss": 0.0046, + "reward": 1.703479528427124, + "reward_std": 0.09505253285169601, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4534795582294464, + "step": 480 + }, + { + "completion_length": 155.15625, + "epoch": 0.1532823454429573, + "grad_norm": 12.259832382202148, + "kl": 0.14453125, + "learning_rate": 8.467176545570427e-07, + "loss": 0.0058, + "reward": 1.3880560398101807, + "reward_std": 0.08335339277982712, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.38805609941482544, + "step": 481 + }, + { + "completion_length": 99.59375, + "epoch": 0.15360101975780752, + "grad_norm": 20.600967407226562, + "kl": 0.1025390625, + "learning_rate": 8.463989802421925e-07, + "loss": 0.0041, + "reward": 1.613484501838684, + "reward_std": 0.05416692793369293, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4884844720363617, + "step": 482 + }, + { + "completion_length": 127.203125, + "epoch": 0.15391969407265774, + "grad_norm": 53.93398666381836, + "kl": 0.05517578125, + "learning_rate": 8.460803059273422e-07, + "loss": 0.0022, + "reward": 1.697056531906128, + "reward_std": 0.09304764866828918, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5720565319061279, + "step": 483 + }, + { + "completion_length": 181.8125, + "epoch": 0.15423836838750796, + "grad_norm": 48.73480987548828, + "kl": 0.052490234375, + "learning_rate": 8.45761631612492e-07, + "loss": 0.0021, + "reward": 1.4899718761444092, + "reward_std": 0.025554783642292023, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.36497190594673157, + "step": 484 + }, + { + "completion_length": 129.859375, + "epoch": 0.15455704270235818, + "grad_norm": 30.73780059814453, + "kl": 0.07470703125, + "learning_rate": 8.454429572976418e-07, + "loss": 0.003, + "reward": 1.611334204673767, + "reward_std": 0.10449053347110748, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.37695926427841187, + "rewards/pad": 0.234375, + "step": 485 + }, + { + "completion_length": 100.640625, + "epoch": 0.15487571701720843, + "grad_norm": 57.61213684082031, + "kl": 0.1435546875, + "learning_rate": 8.451242829827916e-07, + "loss": 0.0057, + "reward": 1.4112203121185303, + "reward_std": 0.16161048412322998, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3799702823162079, + "rewards/pad": 0.03125, + "step": 486 + }, + { + "completion_length": 154.90625, + "epoch": 0.15519439133205865, + "grad_norm": 16.694082260131836, + "kl": 0.095703125, + "learning_rate": 8.448056086679413e-07, + "loss": 0.0038, + "reward": 1.5090348720550537, + "reward_std": 0.040350381284952164, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5090348124504089, + "step": 487 + }, + { + "completion_length": 126.96875, + "epoch": 0.15551306564690887, + "grad_norm": 17.563032150268555, + "kl": 0.1181640625, + "learning_rate": 8.444869343530911e-07, + "loss": 0.0047, + "reward": 1.538849115371704, + "reward_std": 0.09288837015628815, + "rewards/pad": 0.234375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.30447399616241455, + "step": 488 + }, + { + "completion_length": 45.3125, + "epoch": 0.15583173996175909, + "grad_norm": 19.666879653930664, + "kl": 0.181640625, + "learning_rate": 8.44168260038241e-07, + "loss": 0.0073, + "reward": 1.4959934949874878, + "reward_std": 0.151860311627388, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.49599355459213257, + "rewards/pad": 0.0, + "step": 489 + }, + { + "completion_length": 73.28125, + "epoch": 0.1561504142766093, + "grad_norm": 33.24127960205078, + "kl": 0.259765625, + "learning_rate": 8.438495857233908e-07, + "loss": 0.0104, + "reward": 1.5356709957122803, + "reward_std": 0.11758964508771896, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5356709361076355, + "step": 490 + }, + { + "completion_length": 152.703125, + "epoch": 0.15646908859145953, + "grad_norm": 29.862022399902344, + "kl": 0.30859375, + "learning_rate": 8.435309114085404e-07, + "loss": 0.0123, + "reward": 1.3644418716430664, + "reward_std": 0.02914932370185852, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3644419312477112, + "rewards/pad": 0.0, + "step": 491 + }, + { + "completion_length": 100.78125, + "epoch": 0.15678776290630975, + "grad_norm": 147.37142944335938, + "kl": 0.0986328125, + "learning_rate": 8.432122370936902e-07, + "loss": 0.004, + "reward": 1.4690436124801636, + "reward_std": 0.11149085313081741, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3752935826778412, + "rewards/pad": 0.09375, + "step": 492 + }, + { + "completion_length": 46.59375, + "epoch": 0.15710643722115997, + "grad_norm": 169.15550231933594, + "kl": 0.1357421875, + "learning_rate": 8.4289356277884e-07, + "loss": 0.0054, + "reward": 1.700716495513916, + "reward_std": 0.11945337802171707, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5757165551185608, + "step": 493 + }, + { + "completion_length": 72.765625, + "epoch": 0.1574251115360102, + "grad_norm": 58.808162689208984, + "kl": 0.1904296875, + "learning_rate": 8.425748884639897e-07, + "loss": 0.0076, + "reward": 1.3045260906219482, + "reward_std": 0.1511131078004837, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.30452603101730347, + "step": 494 + }, + { + "completion_length": 47.03125, + "epoch": 0.1577437858508604, + "grad_norm": 43.41145706176758, + "kl": 0.1767578125, + "learning_rate": 8.422562141491395e-07, + "loss": 0.0071, + "reward": 1.5499234199523926, + "reward_std": 0.07011061161756516, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4249233901500702, + "rewards/pad": 0.125, + "step": 495 + }, + { + "completion_length": 98.609375, + "epoch": 0.15806246016571066, + "grad_norm": 45.90449905395508, + "kl": 0.09765625, + "learning_rate": 8.419375398342893e-07, + "loss": 0.0039, + "reward": 1.5362162590026855, + "reward_std": 0.08368147164583206, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5362161993980408, + "rewards/pad": 0.0, + "step": 496 + }, + { + "completion_length": 125.5, + "epoch": 0.15838113448056088, + "grad_norm": 26.423377990722656, + "kl": 0.1240234375, + "learning_rate": 8.416188655194391e-07, + "loss": 0.005, + "reward": 1.3670811653137207, + "reward_std": 0.13680918514728546, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.3827061653137207, + "step": 497 + }, + { + "completion_length": 72.828125, + "epoch": 0.1586998087954111, + "grad_norm": 27.449508666992188, + "kl": 0.28515625, + "learning_rate": 8.413001912045888e-07, + "loss": 0.0114, + "reward": 1.7771075963974, + "reward_std": 0.08483612537384033, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5271075963973999, + "rewards/pad": 0.25, + "step": 498 + }, + { + "completion_length": 47.578125, + "epoch": 0.15901848311026132, + "grad_norm": 22.990766525268555, + "kl": 0.142578125, + "learning_rate": 8.409815168897386e-07, + "loss": 0.0057, + "reward": 1.6906874179840088, + "reward_std": 0.07498195022344589, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4406873881816864, + "rewards/pad": 0.25, + "step": 499 + }, + { + "completion_length": 152.90625, + "epoch": 0.15933715742511154, + "grad_norm": 13.581183433532715, + "kl": 0.07666015625, + "learning_rate": 8.406628425748884e-07, + "loss": 0.0031, + "reward": 1.4057791233062744, + "reward_std": 0.03723069280385971, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.405779093503952, + "step": 500 + }, + { + "completion_length": 98.0, + "epoch": 0.15965583173996176, + "grad_norm": 29.467369079589844, + "kl": 0.0927734375, + "learning_rate": 8.403441682600382e-07, + "loss": 0.0037, + "reward": 1.6264564990997314, + "reward_std": 0.1466260701417923, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 0.984375, + "rewards/iou_glue_reward": 0.5170813798904419, + "step": 501 + }, + { + "completion_length": 125.875, + "epoch": 0.15997450605481198, + "grad_norm": 75.3805923461914, + "kl": 0.5625, + "learning_rate": 8.400254939451879e-07, + "loss": 0.0224, + "reward": 1.4467419385910034, + "reward_std": 0.17177796363830566, + "rewards/answer_reward": 0.078125, + "rewards/format_reward_gqa": 0.984375, + "rewards/iou_glue_reward": 0.3842419385910034, + "step": 502 + }, + { + "completion_length": 98.953125, + "epoch": 0.1602931803696622, + "grad_norm": 17.550413131713867, + "kl": 0.1767578125, + "learning_rate": 8.397068196303377e-07, + "loss": 0.0071, + "reward": 1.3917758464813232, + "reward_std": 0.090950608253479, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.39177584648132324, + "step": 503 + }, + { + "completion_length": 124.84375, + "epoch": 0.16061185468451242, + "grad_norm": 18.48185920715332, + "kl": 0.0888671875, + "learning_rate": 8.393881453154875e-07, + "loss": 0.0036, + "reward": 1.5010762214660645, + "reward_std": 0.1343594491481781, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3917011618614197, + "rewards/pad": 0.109375, + "step": 504 + }, + { + "completion_length": 175.859375, + "epoch": 0.16093052899936264, + "grad_norm": 10.59714412689209, + "kl": 0.06494140625, + "learning_rate": 8.390694710006373e-07, + "loss": 0.0026, + "reward": 1.4233787059783936, + "reward_std": 0.051402267068624496, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4233787953853607, + "step": 505 + }, + { + "completion_length": 155.640625, + "epoch": 0.16124920331421289, + "grad_norm": 11.212116241455078, + "kl": 0.06640625, + "learning_rate": 8.38750796685787e-07, + "loss": 0.0026, + "reward": 1.450839877128601, + "reward_std": 0.039273642003536224, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3258398175239563, + "step": 506 + }, + { + "completion_length": 70.5, + "epoch": 0.1615678776290631, + "grad_norm": 19.700626373291016, + "kl": 0.1455078125, + "learning_rate": 8.384321223709368e-07, + "loss": 0.0058, + "reward": 1.5653252601623535, + "reward_std": 0.13670827448368073, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.44032523036003113, + "rewards/pad": 0.125, + "step": 507 + }, + { + "completion_length": 98.8125, + "epoch": 0.16188655194391333, + "grad_norm": 24.49850845336914, + "kl": 0.07080078125, + "learning_rate": 8.381134480560866e-07, + "loss": 0.0028, + "reward": 1.6112509965896606, + "reward_std": 0.12817604839801788, + "rewards/pad": 0.21875, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.39250099658966064, + "step": 508 + }, + { + "completion_length": 132.0625, + "epoch": 0.16220522625876355, + "grad_norm": 15.46005916595459, + "kl": 0.0615234375, + "learning_rate": 8.377947737412365e-07, + "loss": 0.0025, + "reward": 1.5222032070159912, + "reward_std": 0.05216310918331146, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.397203266620636, + "step": 509 + }, + { + "completion_length": 179.265625, + "epoch": 0.16252390057361377, + "grad_norm": 8.991511344909668, + "kl": 0.039794921875, + "learning_rate": 8.374760994263862e-07, + "loss": 0.0016, + "reward": 1.5481150150299072, + "reward_std": 0.05589091405272484, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.42311498522758484, + "step": 510 + }, + { + "completion_length": 102.59375, + "epoch": 0.162842574888464, + "grad_norm": 60.91361618041992, + "kl": 0.10400390625, + "learning_rate": 8.37157425111536e-07, + "loss": 0.0042, + "reward": 1.682870864868164, + "reward_std": 0.07718949019908905, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5578707456588745, + "rewards/pad": 0.125, + "step": 511 + }, + { + "completion_length": 176.59375, + "epoch": 0.1631612492033142, + "grad_norm": 8.942208290100098, + "kl": 0.039794921875, + "learning_rate": 8.368387507966858e-07, + "loss": 0.0016, + "reward": 1.4696202278137207, + "reward_std": 0.033947329968214035, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4696202874183655, + "step": 512 + }, + { + "completion_length": 124.28125, + "epoch": 0.16347992351816443, + "grad_norm": 21.55902862548828, + "kl": 0.353515625, + "learning_rate": 8.365200764818356e-07, + "loss": 0.0141, + "reward": 1.493517518043518, + "reward_std": 0.057740386575460434, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.49351754784584045, + "rewards/pad": 0.0, + "step": 513 + }, + { + "completion_length": 71.59375, + "epoch": 0.16379859783301465, + "grad_norm": 17.975238800048828, + "kl": 0.10693359375, + "learning_rate": 8.362014021669853e-07, + "loss": 0.0043, + "reward": 1.531064510345459, + "reward_std": 0.18845269083976746, + "rewards/answer_reward": 0.09375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4373144805431366, + "step": 514 + }, + { + "completion_length": 45.078125, + "epoch": 0.1641172721478649, + "grad_norm": 19.34892463684082, + "kl": 0.12890625, + "learning_rate": 8.358827278521351e-07, + "loss": 0.0052, + "reward": 1.394120216369629, + "reward_std": 0.09514320641756058, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3941202461719513, + "rewards/pad": 0.0, + "step": 515 + }, + { + "completion_length": 99.84375, + "epoch": 0.16443594646271512, + "grad_norm": 43.57005310058594, + "kl": 0.0859375, + "learning_rate": 8.355640535372849e-07, + "loss": 0.0034, + "reward": 1.5611262321472168, + "reward_std": 0.10545508563518524, + "rewards/pad": 0.046875, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5142512321472168, + "step": 516 + }, + { + "completion_length": 99.75, + "epoch": 0.16475462077756534, + "grad_norm": 34.81675720214844, + "kl": 0.08837890625, + "learning_rate": 8.352453792224347e-07, + "loss": 0.0035, + "reward": 1.697974681854248, + "reward_std": 0.14656002819538116, + "rewards/pad": 0.171875, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.526099681854248, + "step": 517 + }, + { + "completion_length": 130.859375, + "epoch": 0.16507329509241556, + "grad_norm": 20.660293579101562, + "kl": 0.0791015625, + "learning_rate": 8.349267049075844e-07, + "loss": 0.0032, + "reward": 1.4722533226013184, + "reward_std": 0.07390671223402023, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.34725329279899597, + "rewards/pad": 0.125, + "step": 518 + }, + { + "completion_length": 152.78125, + "epoch": 0.16539196940726578, + "grad_norm": 21.820556640625, + "kl": 0.08203125, + "learning_rate": 8.346080305927342e-07, + "loss": 0.0033, + "reward": 1.314011573791504, + "reward_std": 0.04784318059682846, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3140115737915039, + "step": 519 + }, + { + "completion_length": 45.515625, + "epoch": 0.165710643722116, + "grad_norm": 43.039615631103516, + "kl": 0.1259765625, + "learning_rate": 8.34289356277884e-07, + "loss": 0.0051, + "reward": 1.6405762434005737, + "reward_std": 0.16814163327217102, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5937012434005737, + "rewards/pad": 0.046875, + "step": 520 + }, + { + "completion_length": 122.75, + "epoch": 0.16602931803696622, + "grad_norm": 14.376192092895508, + "kl": 0.1298828125, + "learning_rate": 8.339706819630338e-07, + "loss": 0.0052, + "reward": 1.4471235275268555, + "reward_std": 0.06147985905408859, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.44712358713150024, + "step": 521 + }, + { + "completion_length": 98.84375, + "epoch": 0.16634799235181644, + "grad_norm": 16.817852020263672, + "kl": 0.11962890625, + "learning_rate": 8.336520076481835e-07, + "loss": 0.0048, + "reward": 1.396544098854065, + "reward_std": 0.14501231908798218, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.41216909885406494, + "rewards/pad": 0.0, + "step": 522 + }, + { + "completion_length": 94.125, + "epoch": 0.16666666666666666, + "grad_norm": 16.053434371948242, + "kl": 0.1279296875, + "learning_rate": 8.333333333333333e-07, + "loss": 0.0051, + "reward": 1.5878493785858154, + "reward_std": 0.17199000716209412, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.47847428917884827, + "rewards/pad": 0.125, + "step": 523 + }, + { + "completion_length": 73.546875, + "epoch": 0.16698534098151688, + "grad_norm": 63.2349853515625, + "kl": 0.1484375, + "learning_rate": 8.330146590184831e-07, + "loss": 0.0059, + "reward": 1.5571330785751343, + "reward_std": 0.18380963802337646, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.43213310837745667, + "rewards/pad": 0.125, + "step": 524 + }, + { + "completion_length": 185.515625, + "epoch": 0.16730401529636713, + "grad_norm": 13.684309959411621, + "kl": 0.06689453125, + "learning_rate": 8.326959847036329e-07, + "loss": 0.0027, + "reward": 1.6064646244049072, + "reward_std": 0.05707468092441559, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.481464684009552, + "step": 525 + }, + { + "completion_length": 100.328125, + "epoch": 0.16762268961121735, + "grad_norm": 20.025402069091797, + "kl": 0.078125, + "learning_rate": 8.323773103887826e-07, + "loss": 0.0031, + "reward": 1.5280306339263916, + "reward_std": 0.07771054655313492, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.40303054451942444, + "step": 526 + }, + { + "completion_length": 72.96875, + "epoch": 0.16794136392606757, + "grad_norm": 20.100772857666016, + "kl": 0.169921875, + "learning_rate": 8.320586360739325e-07, + "loss": 0.0068, + "reward": 1.463815450668335, + "reward_std": 0.14635515213012695, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.4794403910636902, + "rewards/pad": 0.0, + "step": 527 + }, + { + "completion_length": 96.71875, + "epoch": 0.1682600382409178, + "grad_norm": 60.14537048339844, + "kl": 0.10791015625, + "learning_rate": 8.317399617590823e-07, + "loss": 0.0043, + "reward": 1.6089813709259033, + "reward_std": 0.06876353919506073, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.6089814305305481, + "step": 528 + }, + { + "completion_length": 124.0, + "epoch": 0.168578712555768, + "grad_norm": 41.38592529296875, + "kl": 0.12158203125, + "learning_rate": 8.31421287444232e-07, + "loss": 0.0049, + "reward": 1.5476176738739014, + "reward_std": 0.08636157214641571, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.42261767387390137, + "rewards/pad": 0.125, + "step": 529 + }, + { + "completion_length": 123.09375, + "epoch": 0.16889738687061823, + "grad_norm": 21.460275650024414, + "kl": 0.11328125, + "learning_rate": 8.311026131293817e-07, + "loss": 0.0045, + "reward": 1.5313866138458252, + "reward_std": 0.07336382567882538, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5313864946365356, + "rewards/pad": 0.0, + "step": 530 + }, + { + "completion_length": 97.375, + "epoch": 0.16921606118546845, + "grad_norm": 61.537071228027344, + "kl": 0.1044921875, + "learning_rate": 8.307839388145315e-07, + "loss": 0.0042, + "reward": 1.7152791023254395, + "reward_std": 0.1346595138311386, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.48090416193008423, + "rewards/pad": 0.234375, + "step": 531 + }, + { + "completion_length": 124.046875, + "epoch": 0.16953473550031867, + "grad_norm": 119.3514633178711, + "kl": 0.0791015625, + "learning_rate": 8.304652644996813e-07, + "loss": 0.0032, + "reward": 1.3922349214553833, + "reward_std": 0.0685240775346756, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3922349512577057, + "rewards/pad": 0.0, + "step": 532 + }, + { + "completion_length": 47.28125, + "epoch": 0.1698534098151689, + "grad_norm": 35.74647903442383, + "kl": 0.1572265625, + "learning_rate": 8.30146590184831e-07, + "loss": 0.0063, + "reward": 1.6910107135772705, + "reward_std": 0.17851702868938446, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.36288559436798096, + "rewards/pad": 0.328125, + "step": 533 + }, + { + "completion_length": 97.734375, + "epoch": 0.1701720841300191, + "grad_norm": 10.062421798706055, + "kl": 0.1298828125, + "learning_rate": 8.298279158699808e-07, + "loss": 0.0052, + "reward": 1.5344090461730957, + "reward_std": 0.06182171404361725, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4094090759754181, + "rewards/pad": 0.125, + "step": 534 + }, + { + "completion_length": 98.390625, + "epoch": 0.17049075844486936, + "grad_norm": 22.848539352416992, + "kl": 0.20703125, + "learning_rate": 8.295092415551306e-07, + "loss": 0.0083, + "reward": 1.459669828414917, + "reward_std": 0.08608236908912659, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.459669828414917, + "step": 535 + }, + { + "completion_length": 99.671875, + "epoch": 0.17080943275971958, + "grad_norm": 32.05093765258789, + "kl": 0.09521484375, + "learning_rate": 8.291905672402804e-07, + "loss": 0.0038, + "reward": 1.8032017946243286, + "reward_std": 0.08446711301803589, + "rewards/pad": 0.375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.42820173501968384, + "step": 536 + }, + { + "completion_length": 21.140625, + "epoch": 0.1711281070745698, + "grad_norm": 44.99974060058594, + "kl": 0.1337890625, + "learning_rate": 8.288718929254301e-07, + "loss": 0.0054, + "reward": 1.8819546699523926, + "reward_std": 0.10786371678113937, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5069546699523926, + "rewards/pad": 0.375, + "step": 537 + }, + { + "completion_length": 73.453125, + "epoch": 0.17144678138942002, + "grad_norm": 41.52457046508789, + "kl": 0.08642578125, + "learning_rate": 8.285532186105799e-07, + "loss": 0.0034, + "reward": 1.7312819957733154, + "reward_std": 0.12865957617759705, + "rewards/pad": 0.359375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3719070553779602, + "step": 538 + }, + { + "completion_length": 124.828125, + "epoch": 0.17176545570427024, + "grad_norm": 34.48262023925781, + "kl": 0.10546875, + "learning_rate": 8.282345442957297e-07, + "loss": 0.0042, + "reward": 1.6211336851119995, + "reward_std": 0.11349187791347504, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4961336851119995, + "rewards/pad": 0.125, + "step": 539 + }, + { + "completion_length": 122.203125, + "epoch": 0.17208413001912046, + "grad_norm": 37.691978454589844, + "kl": 0.0673828125, + "learning_rate": 8.279158699808795e-07, + "loss": 0.0027, + "reward": 1.7140142917633057, + "reward_std": 0.0915536880493164, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4640142321586609, + "rewards/pad": 0.25, + "step": 540 + }, + { + "completion_length": 149.6875, + "epoch": 0.17240280433397068, + "grad_norm": 42.488380432128906, + "kl": 0.08349609375, + "learning_rate": 8.275971956660292e-07, + "loss": 0.0033, + "reward": 1.5138425827026367, + "reward_std": 0.07240451872348785, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5138426423072815, + "step": 541 + }, + { + "completion_length": 95.84375, + "epoch": 0.1727214786488209, + "grad_norm": 669.1302490234375, + "kl": 0.1279296875, + "learning_rate": 8.27278521351179e-07, + "loss": 0.0051, + "reward": 1.321406602859497, + "reward_std": 0.10743975639343262, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3214065134525299, + "rewards/pad": 0.0, + "step": 542 + }, + { + "completion_length": 97.609375, + "epoch": 0.17304015296367112, + "grad_norm": 37.46754455566406, + "kl": 0.08544921875, + "learning_rate": 8.269598470363288e-07, + "loss": 0.0034, + "reward": 1.5845844745635986, + "reward_std": 0.11835047602653503, + "rewards/answer_reward": 0.15625, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.42833447456359863, + "step": 543 + }, + { + "completion_length": 71.609375, + "epoch": 0.17335882727852134, + "grad_norm": 83.38258361816406, + "kl": 0.1259765625, + "learning_rate": 8.266411727214786e-07, + "loss": 0.005, + "reward": 1.4551091194152832, + "reward_std": 0.11183137446641922, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3769841194152832, + "rewards/pad": 0.078125, + "step": 544 + }, + { + "completion_length": 97.40625, + "epoch": 0.17367750159337159, + "grad_norm": 19.901351928710938, + "kl": 0.1533203125, + "learning_rate": 8.263224984066283e-07, + "loss": 0.0061, + "reward": 1.5000262260437012, + "reward_std": 0.10315240919589996, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5000262260437012, + "rewards/pad": 0.0, + "step": 545 + }, + { + "completion_length": 45.46875, + "epoch": 0.1739961759082218, + "grad_norm": 21.25617218017578, + "kl": 0.12060546875, + "learning_rate": 8.260038240917782e-07, + "loss": 0.0048, + "reward": 1.5835143327713013, + "reward_std": 0.09237085282802582, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5835143327713013, + "rewards/pad": 0.0, + "step": 546 + }, + { + "completion_length": 98.421875, + "epoch": 0.17431485022307203, + "grad_norm": 25.097929000854492, + "kl": 0.08837890625, + "learning_rate": 8.25685149776928e-07, + "loss": 0.0035, + "reward": 1.6923954486846924, + "reward_std": 0.04992898926138878, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5673954486846924, + "step": 547 + }, + { + "completion_length": 97.890625, + "epoch": 0.17463352453792225, + "grad_norm": 97.4831771850586, + "kl": 0.109375, + "learning_rate": 8.253664754620778e-07, + "loss": 0.0044, + "reward": 1.6336106061935425, + "reward_std": 0.1055816188454628, + "rewards/pad": 0.109375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5242354869842529, + "step": 548 + }, + { + "completion_length": 122.4375, + "epoch": 0.17495219885277247, + "grad_norm": 27.211618423461914, + "kl": 0.1337890625, + "learning_rate": 8.250478011472275e-07, + "loss": 0.0054, + "reward": 1.476280927658081, + "reward_std": 0.05999341607093811, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.47628092765808105, + "step": 549 + }, + { + "completion_length": 47.625, + "epoch": 0.1752708731676227, + "grad_norm": 62.962547302246094, + "kl": 0.126953125, + "learning_rate": 8.247291268323773e-07, + "loss": 0.0051, + "reward": 1.7597732543945312, + "reward_std": 0.23647846281528473, + "rewards/answer_reward": 0.359375, + "rewards/format_reward_gqa": 0.984375, + "rewards/iou_glue_reward": 0.4160231053829193, + "step": 550 + }, + { + "completion_length": 123.65625, + "epoch": 0.1755895474824729, + "grad_norm": 40.25691604614258, + "kl": 0.07177734375, + "learning_rate": 8.244104525175271e-07, + "loss": 0.0029, + "reward": 1.5125081539154053, + "reward_std": 0.10185742378234863, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.52813321352005, + "step": 551 + }, + { + "completion_length": 98.796875, + "epoch": 0.17590822179732313, + "grad_norm": 27.152267456054688, + "kl": 0.11572265625, + "learning_rate": 8.240917782026769e-07, + "loss": 0.0046, + "reward": 1.339756965637207, + "reward_std": 0.06058652698993683, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3397570252418518, + "step": 552 + }, + { + "completion_length": 175.9375, + "epoch": 0.17622689611217335, + "grad_norm": 10.05562686920166, + "kl": 0.09521484375, + "learning_rate": 8.237731038878266e-07, + "loss": 0.0038, + "reward": 1.601670265197754, + "reward_std": 0.04825048893690109, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4766702950000763, + "rewards/pad": 0.125, + "step": 553 + }, + { + "completion_length": 69.734375, + "epoch": 0.17654557042702357, + "grad_norm": 71.13457489013672, + "kl": 0.1494140625, + "learning_rate": 8.234544295729764e-07, + "loss": 0.006, + "reward": 1.588582158088684, + "reward_std": 0.12747931480407715, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4792071580886841, + "rewards/pad": 0.109375, + "step": 554 + }, + { + "completion_length": 70.953125, + "epoch": 0.17686424474187382, + "grad_norm": 23.054285049438477, + "kl": 0.1806640625, + "learning_rate": 8.231357552581262e-07, + "loss": 0.0072, + "reward": 1.4141204357147217, + "reward_std": 0.14130906760692596, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4141203463077545, + "rewards/pad": 0.0, + "step": 555 + }, + { + "completion_length": 123.109375, + "epoch": 0.17718291905672404, + "grad_norm": 24.595977783203125, + "kl": 0.08544921875, + "learning_rate": 8.228170809432759e-07, + "loss": 0.0034, + "reward": 1.548499584197998, + "reward_std": 0.052562348544597626, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5484994649887085, + "step": 556 + }, + { + "completion_length": 126.328125, + "epoch": 0.17750159337157426, + "grad_norm": 42.06749725341797, + "kl": 0.0751953125, + "learning_rate": 8.224984066284257e-07, + "loss": 0.003, + "reward": 1.403905987739563, + "reward_std": 0.08751250803470612, + "rewards/answer_reward": 0.015625, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.3882809579372406, + "step": 557 + }, + { + "completion_length": 71.40625, + "epoch": 0.17782026768642448, + "grad_norm": 27.221906661987305, + "kl": 0.2265625, + "learning_rate": 8.221797323135755e-07, + "loss": 0.0091, + "reward": 1.4237648248672485, + "reward_std": 0.0997081995010376, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4237648844718933, + "rewards/pad": 0.0, + "step": 558 + }, + { + "completion_length": 44.96875, + "epoch": 0.1781389420012747, + "grad_norm": 16.742216110229492, + "kl": 0.1630859375, + "learning_rate": 8.218610579987253e-07, + "loss": 0.0065, + "reward": 1.3757002353668213, + "reward_std": 0.09360139071941376, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3757002055644989, + "rewards/pad": 0.0, + "step": 559 + }, + { + "completion_length": 147.6875, + "epoch": 0.17845761631612492, + "grad_norm": 12.943758010864258, + "kl": 0.07958984375, + "learning_rate": 8.21542383683875e-07, + "loss": 0.0032, + "reward": 1.3315762281417847, + "reward_std": 0.039429426193237305, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.33157628774642944, + "step": 560 + }, + { + "completion_length": 71.5625, + "epoch": 0.17877629063097514, + "grad_norm": 75.57307434082031, + "kl": 0.23828125, + "learning_rate": 8.212237093690248e-07, + "loss": 0.0095, + "reward": 1.5689356327056885, + "reward_std": 0.10561536252498627, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.568935751914978, + "rewards/pad": 0.0, + "step": 561 + }, + { + "completion_length": 71.671875, + "epoch": 0.17909496494582536, + "grad_norm": 26.024625778198242, + "kl": 0.1953125, + "learning_rate": 8.209050350541746e-07, + "loss": 0.0078, + "reward": 1.5372991561889648, + "reward_std": 0.09807107597589493, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5372991561889648, + "step": 562 + }, + { + "completion_length": 71.421875, + "epoch": 0.17941363926067558, + "grad_norm": 28.438175201416016, + "kl": 0.189453125, + "learning_rate": 8.205863607393244e-07, + "loss": 0.0076, + "reward": 1.480418086051941, + "reward_std": 0.07410193979740143, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3554180860519409, + "rewards/pad": 0.125, + "step": 563 + }, + { + "completion_length": 96.1875, + "epoch": 0.17973231357552583, + "grad_norm": 47.855194091796875, + "kl": 0.1435546875, + "learning_rate": 8.202676864244741e-07, + "loss": 0.0058, + "reward": 1.385023593902588, + "reward_std": 0.13159047067165375, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3225235939025879, + "rewards/pad": 0.0625, + "step": 564 + }, + { + "completion_length": 97.8125, + "epoch": 0.18005098789037605, + "grad_norm": 17.224849700927734, + "kl": 0.1611328125, + "learning_rate": 8.19949012109624e-07, + "loss": 0.0064, + "reward": 1.4678709506988525, + "reward_std": 0.061766088008880615, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.46787095069885254, + "step": 565 + }, + { + "completion_length": 120.28125, + "epoch": 0.18036966220522627, + "grad_norm": 14.757925033569336, + "kl": 0.10009765625, + "learning_rate": 8.196303377947738e-07, + "loss": 0.004, + "reward": 1.3626590967178345, + "reward_std": 0.05897171050310135, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3626590967178345, + "step": 566 + }, + { + "completion_length": 71.03125, + "epoch": 0.1806883365200765, + "grad_norm": 220.81106567382812, + "kl": 0.1337890625, + "learning_rate": 8.193116634799236e-07, + "loss": 0.0054, + "reward": 1.5400254726409912, + "reward_std": 0.11256685853004456, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4150254726409912, + "rewards/pad": 0.125, + "step": 567 + }, + { + "completion_length": 125.609375, + "epoch": 0.1810070108349267, + "grad_norm": 603.3175659179688, + "kl": 0.1005859375, + "learning_rate": 8.189929891650733e-07, + "loss": 0.004, + "reward": 1.4392961263656616, + "reward_std": 0.10471741855144501, + "rewards/pad": 0.015625, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4236711263656616, + "step": 568 + }, + { + "completion_length": 150.609375, + "epoch": 0.18132568514977693, + "grad_norm": 216.56732177734375, + "kl": 0.08154296875, + "learning_rate": 8.18674314850223e-07, + "loss": 0.0033, + "reward": 1.5750045776367188, + "reward_std": 0.09949097037315369, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.49687960743904114, + "rewards/pad": 0.078125, + "step": 569 + }, + { + "completion_length": 99.515625, + "epoch": 0.18164435946462715, + "grad_norm": 43.249183654785156, + "kl": 0.08837890625, + "learning_rate": 8.183556405353728e-07, + "loss": 0.0035, + "reward": 1.7908916473388672, + "reward_std": 0.0621541291475296, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.41589170694351196, + "rewards/pad": 0.375, + "step": 570 + }, + { + "completion_length": 148.765625, + "epoch": 0.18196303377947737, + "grad_norm": 37.219024658203125, + "kl": 0.125, + "learning_rate": 8.180369662205226e-07, + "loss": 0.005, + "reward": 1.535273551940918, + "reward_std": 0.03435523062944412, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.535273551940918, + "rewards/pad": 0.0, + "step": 571 + }, + { + "completion_length": 70.875, + "epoch": 0.1822817080943276, + "grad_norm": 28.961162567138672, + "kl": 0.10205078125, + "learning_rate": 8.177182919056723e-07, + "loss": 0.0041, + "reward": 1.67730712890625, + "reward_std": 0.13280938565731049, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 0.984375, + "rewards/iou_glue_reward": 0.44293221831321716, + "step": 572 + }, + { + "completion_length": 123.71875, + "epoch": 0.1826003824091778, + "grad_norm": 43.969078063964844, + "kl": 0.166015625, + "learning_rate": 8.173996175908221e-07, + "loss": 0.0066, + "reward": 1.3603620529174805, + "reward_std": 0.07749934494495392, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3603619933128357, + "rewards/pad": 0.0, + "step": 573 + }, + { + "completion_length": 123.6875, + "epoch": 0.18291905672402806, + "grad_norm": 56.06167221069336, + "kl": 0.09619140625, + "learning_rate": 8.170809432759719e-07, + "loss": 0.0039, + "reward": 1.6252659559249878, + "reward_std": 0.06589239835739136, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.500265896320343, + "rewards/pad": 0.125, + "step": 574 + }, + { + "completion_length": 98.546875, + "epoch": 0.18323773103887828, + "grad_norm": 28.925315856933594, + "kl": 0.1630859375, + "learning_rate": 8.167622689611217e-07, + "loss": 0.0065, + "reward": 1.5602610111236572, + "reward_std": 0.07640358805656433, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4352610111236572, + "rewards/pad": 0.125, + "step": 575 + }, + { + "completion_length": 45.59375, + "epoch": 0.1835564053537285, + "grad_norm": 62.40200424194336, + "kl": 0.42578125, + "learning_rate": 8.164435946462714e-07, + "loss": 0.017, + "reward": 1.6693676710128784, + "reward_std": 0.08748309314250946, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4193676710128784, + "rewards/pad": 0.25, + "step": 576 + }, + { + "completion_length": 70.6875, + "epoch": 0.18387507966857872, + "grad_norm": 35.13634490966797, + "kl": 0.1953125, + "learning_rate": 8.161249203314212e-07, + "loss": 0.0078, + "reward": 1.6500343084335327, + "reward_std": 0.11489731073379517, + "rewards/pad": 0.234375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4156593084335327, + "step": 577 + }, + { + "completion_length": 123.15625, + "epoch": 0.18419375398342894, + "grad_norm": 52.97418212890625, + "kl": 0.2890625, + "learning_rate": 8.15806246016571e-07, + "loss": 0.0116, + "reward": 1.5006070137023926, + "reward_std": 0.08267797529697418, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5006070733070374, + "step": 578 + }, + { + "completion_length": 46.859375, + "epoch": 0.18451242829827916, + "grad_norm": 94.87614440917969, + "kl": 0.10302734375, + "learning_rate": 8.154875717017208e-07, + "loss": 0.0041, + "reward": 1.465181827545166, + "reward_std": 0.06751959025859833, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.3401819169521332, + "step": 579 + }, + { + "completion_length": 97.046875, + "epoch": 0.18483110261312938, + "grad_norm": 35.05488586425781, + "kl": 0.30078125, + "learning_rate": 8.151688973868705e-07, + "loss": 0.0121, + "reward": 1.5429270267486572, + "reward_std": 0.12125824391841888, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5429270267486572, + "rewards/pad": 0.0, + "step": 580 + }, + { + "completion_length": 48.671875, + "epoch": 0.1851497769279796, + "grad_norm": 88.4456558227539, + "kl": 0.11083984375, + "learning_rate": 8.148502230720203e-07, + "loss": 0.0044, + "reward": 1.610171914100647, + "reward_std": 0.16949138045310974, + "rewards/pad": 0.34375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.26642197370529175, + "step": 581 + }, + { + "completion_length": 172.65625, + "epoch": 0.18546845124282982, + "grad_norm": 19.403085708618164, + "kl": 0.0556640625, + "learning_rate": 8.145315487571701e-07, + "loss": 0.0022, + "reward": 1.3586130142211914, + "reward_std": 0.12332511693239212, + "rewards/answer_reward": 0.015625, + "rewards/format_reward_gqa": 0.984375, + "rewards/iou_glue_reward": 0.358612984418869, + "step": 582 + }, + { + "completion_length": 42.953125, + "epoch": 0.18578712555768004, + "grad_norm": 47.306217193603516, + "kl": 0.119140625, + "learning_rate": 8.1421287444232e-07, + "loss": 0.0048, + "reward": 1.6474767923355103, + "reward_std": 0.057992108166217804, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5224767923355103, + "step": 583 + }, + { + "completion_length": 96.65625, + "epoch": 0.1861057998725303, + "grad_norm": 37.99491500854492, + "kl": 0.0849609375, + "learning_rate": 8.138942001274697e-07, + "loss": 0.0034, + "reward": 1.5028228759765625, + "reward_std": 0.23027420043945312, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 0.96875, + "rewards/iou_glue_reward": 0.4090728163719177, + "step": 584 + }, + { + "completion_length": 176.5625, + "epoch": 0.1864244741873805, + "grad_norm": 17.070524215698242, + "kl": 0.07763671875, + "learning_rate": 8.135755258126195e-07, + "loss": 0.0031, + "reward": 1.446868658065796, + "reward_std": 0.09230832755565643, + "rewards/pad": 0.109375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3374937176704407, + "step": 585 + }, + { + "completion_length": 172.953125, + "epoch": 0.18674314850223073, + "grad_norm": 15.915264129638672, + "kl": 0.048583984375, + "learning_rate": 8.132568514977693e-07, + "loss": 0.0019, + "reward": 1.546690583229065, + "reward_std": 0.09776504337787628, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.43731555342674255, + "step": 586 + }, + { + "completion_length": 150.265625, + "epoch": 0.18706182281708095, + "grad_norm": 18.803260803222656, + "kl": 0.1044921875, + "learning_rate": 8.12938177182919e-07, + "loss": 0.0042, + "reward": 1.3496114015579224, + "reward_std": 0.04225423187017441, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.34961140155792236, + "rewards/pad": 0.0, + "step": 587 + }, + { + "completion_length": 173.234375, + "epoch": 0.18738049713193117, + "grad_norm": 7.521479606628418, + "kl": 0.0771484375, + "learning_rate": 8.126195028680688e-07, + "loss": 0.0031, + "reward": 1.3901759386062622, + "reward_std": 0.13265454769134521, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.40580081939697266, + "rewards/pad": 0.0, + "step": 588 + }, + { + "completion_length": 97.578125, + "epoch": 0.1876991714467814, + "grad_norm": 45.2459716796875, + "kl": 0.6015625, + "learning_rate": 8.123008285532186e-07, + "loss": 0.0242, + "reward": 1.5526204109191895, + "reward_std": 0.08004587888717651, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.42762038111686707, + "rewards/pad": 0.125, + "step": 589 + }, + { + "completion_length": 144.875, + "epoch": 0.1880178457616316, + "grad_norm": 34.44723129272461, + "kl": 0.091796875, + "learning_rate": 8.119821542383684e-07, + "loss": 0.0037, + "reward": 1.4470405578613281, + "reward_std": 0.0640207901597023, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.44704049825668335, + "step": 590 + }, + { + "completion_length": 121.9375, + "epoch": 0.18833652007648183, + "grad_norm": 21.2751407623291, + "kl": 0.11181640625, + "learning_rate": 8.116634799235181e-07, + "loss": 0.0045, + "reward": 1.552978515625, + "reward_std": 0.05177850276231766, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4279784560203552, + "step": 591 + }, + { + "completion_length": 97.953125, + "epoch": 0.18865519439133205, + "grad_norm": 36.41278839111328, + "kl": 0.10888671875, + "learning_rate": 8.113448056086679e-07, + "loss": 0.0044, + "reward": 1.3514432907104492, + "reward_std": 0.08961690962314606, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.35144323110580444, + "rewards/pad": 0.0, + "step": 592 + }, + { + "completion_length": 122.171875, + "epoch": 0.18897386870618227, + "grad_norm": 23.956724166870117, + "kl": 0.1123046875, + "learning_rate": 8.110261312938177e-07, + "loss": 0.0045, + "reward": 1.3557486534118652, + "reward_std": 0.08091498911380768, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3557485342025757, + "step": 593 + }, + { + "completion_length": 123.015625, + "epoch": 0.18929254302103252, + "grad_norm": 73.67423248291016, + "kl": 0.10107421875, + "learning_rate": 8.107074569789675e-07, + "loss": 0.004, + "reward": 1.5595533847808838, + "reward_std": 0.0999167412519455, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5595533847808838, + "step": 594 + }, + { + "completion_length": 45.4375, + "epoch": 0.18961121733588274, + "grad_norm": 18.285587310791016, + "kl": 0.1552734375, + "learning_rate": 8.103887826641172e-07, + "loss": 0.0062, + "reward": 1.662194013595581, + "reward_std": 0.08547796308994293, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.537194013595581, + "rewards/pad": 0.125, + "step": 595 + }, + { + "completion_length": 70.6875, + "epoch": 0.18992989165073296, + "grad_norm": 20.423418045043945, + "kl": 0.1455078125, + "learning_rate": 8.10070108349267e-07, + "loss": 0.0058, + "reward": 1.7276816368103027, + "reward_std": 0.1355665922164917, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6026816368103027, + "rewards/pad": 0.125, + "step": 596 + }, + { + "completion_length": 175.640625, + "epoch": 0.19024856596558318, + "grad_norm": 4.760740280151367, + "kl": 0.09716796875, + "learning_rate": 8.097514340344168e-07, + "loss": 0.0039, + "reward": 1.4160912036895752, + "reward_std": 0.04427863657474518, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4160912334918976, + "step": 597 + }, + { + "completion_length": 45.375, + "epoch": 0.1905672402804334, + "grad_norm": 70.4986343383789, + "kl": 0.1552734375, + "learning_rate": 8.094327597195666e-07, + "loss": 0.0062, + "reward": 1.5294097661972046, + "reward_std": 0.09039345383644104, + "rewards/answer_reward": 0.0, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5294097661972046, + "step": 598 + }, + { + "completion_length": 71.28125, + "epoch": 0.19088591459528362, + "grad_norm": 47.24021530151367, + "kl": 0.130859375, + "learning_rate": 8.091140854047163e-07, + "loss": 0.0052, + "reward": 1.696016550064087, + "reward_std": 0.19736725091934204, + "rewards/pad": 0.296875, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.39914146065711975, + "step": 599 + }, + { + "completion_length": 19.96875, + "epoch": 0.19120458891013384, + "grad_norm": 108.0482177734375, + "kl": 0.244140625, + "learning_rate": 8.087954110898661e-07, + "loss": 0.0098, + "reward": 1.8678321838378906, + "reward_std": 0.18381820619106293, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6959572434425354, + "rewards/pad": 0.171875, + "step": 600 + }, + { + "completion_length": 95.078125, + "epoch": 0.19152326322498406, + "grad_norm": 53.70155715942383, + "kl": 0.216796875, + "learning_rate": 8.084767367750159e-07, + "loss": 0.0087, + "reward": 1.633589744567871, + "reward_std": 0.08066779375076294, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.6335896849632263, + "step": 601 + }, + { + "completion_length": 119.21875, + "epoch": 0.19184193753983428, + "grad_norm": 12.007109642028809, + "kl": 0.09912109375, + "learning_rate": 8.081580624601658e-07, + "loss": 0.004, + "reward": 1.506300687789917, + "reward_std": 0.06164667010307312, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.506300687789917, + "rewards/pad": 0.0, + "step": 602 + }, + { + "completion_length": 94.890625, + "epoch": 0.1921606118546845, + "grad_norm": 19.688268661499023, + "kl": 0.185546875, + "learning_rate": 8.078393881453155e-07, + "loss": 0.0074, + "reward": 1.5328928232192993, + "reward_std": 0.06634339690208435, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5328927040100098, + "rewards/pad": 0.0, + "step": 603 + }, + { + "completion_length": 45.875, + "epoch": 0.19247928616953475, + "grad_norm": 87.2319107055664, + "kl": 0.150390625, + "learning_rate": 8.075207138304653e-07, + "loss": 0.006, + "reward": 1.7554612159729004, + "reward_std": 0.09936299175024033, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5054613351821899, + "rewards/pad": 0.25, + "step": 604 + }, + { + "completion_length": 69.6875, + "epoch": 0.19279796048438497, + "grad_norm": 35.294227600097656, + "kl": 0.232421875, + "learning_rate": 8.072020395156151e-07, + "loss": 0.0093, + "reward": 1.4599926471710205, + "reward_std": 0.04127166420221329, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.33499255776405334, + "step": 605 + }, + { + "completion_length": 72.140625, + "epoch": 0.1931166347992352, + "grad_norm": 43.590309143066406, + "kl": 0.09814453125, + "learning_rate": 8.068833652007649e-07, + "loss": 0.0039, + "reward": 1.6009583473205566, + "reward_std": 0.07119986414909363, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.47595831751823425, + "step": 606 + }, + { + "completion_length": 149.171875, + "epoch": 0.1934353091140854, + "grad_norm": 41.91403579711914, + "kl": 0.0751953125, + "learning_rate": 8.065646908859146e-07, + "loss": 0.003, + "reward": 1.6249721050262451, + "reward_std": 0.05708273500204086, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.49997201561927795, + "step": 607 + }, + { + "completion_length": 94.34375, + "epoch": 0.19375398342893563, + "grad_norm": 22.897380828857422, + "kl": 0.09814453125, + "learning_rate": 8.062460165710643e-07, + "loss": 0.0039, + "reward": 1.4169743061065674, + "reward_std": 0.07337184995412827, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4169743061065674, + "step": 608 + }, + { + "completion_length": 71.609375, + "epoch": 0.19407265774378585, + "grad_norm": 37.60327911376953, + "kl": 0.2255859375, + "learning_rate": 8.059273422562141e-07, + "loss": 0.009, + "reward": 1.5804510116577148, + "reward_std": 0.11089134961366653, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4554510712623596, + "step": 609 + }, + { + "completion_length": 70.875, + "epoch": 0.19439133205863607, + "grad_norm": 27.8912296295166, + "kl": 0.1640625, + "learning_rate": 8.056086679413639e-07, + "loss": 0.0066, + "reward": 1.7579829692840576, + "reward_std": 0.07795955240726471, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6329829692840576, + "rewards/pad": 0.125, + "step": 610 + }, + { + "completion_length": 119.65625, + "epoch": 0.1947100063734863, + "grad_norm": 22.978052139282227, + "kl": 0.1708984375, + "learning_rate": 8.052899936265136e-07, + "loss": 0.0068, + "reward": 1.6393288373947144, + "reward_std": 0.06025080382823944, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5143288969993591, + "rewards/pad": 0.125, + "step": 611 + }, + { + "completion_length": 91.90625, + "epoch": 0.1950286806883365, + "grad_norm": 61.15570068359375, + "kl": 0.1474609375, + "learning_rate": 8.049713193116634e-07, + "loss": 0.0059, + "reward": 1.4246989488601685, + "reward_std": 0.0907503068447113, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.42469894886016846, + "rewards/pad": 0.0, + "step": 612 + }, + { + "completion_length": 70.859375, + "epoch": 0.19534735500318676, + "grad_norm": 18.17074203491211, + "kl": 0.130859375, + "learning_rate": 8.046526449968132e-07, + "loss": 0.0052, + "reward": 1.5264906883239746, + "reward_std": 0.09300049394369125, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5264906883239746, + "rewards/pad": 0.0, + "step": 613 + }, + { + "completion_length": 68.375, + "epoch": 0.19566602931803698, + "grad_norm": 44.83185577392578, + "kl": 0.1572265625, + "learning_rate": 8.043339706819629e-07, + "loss": 0.0063, + "reward": 1.7386105060577393, + "reward_std": 0.10457254946231842, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.48861050605773926, + "step": 614 + }, + { + "completion_length": 70.609375, + "epoch": 0.1959847036328872, + "grad_norm": 13.184401512145996, + "kl": 0.130859375, + "learning_rate": 8.040152963671127e-07, + "loss": 0.0052, + "reward": 1.5657302141189575, + "reward_std": 0.09259041398763657, + "rewards/pad": 0.234375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3313552141189575, + "step": 615 + }, + { + "completion_length": 71.203125, + "epoch": 0.19630337794773742, + "grad_norm": 20.05096435546875, + "kl": 0.1611328125, + "learning_rate": 8.036966220522625e-07, + "loss": 0.0064, + "reward": 1.461806058883667, + "reward_std": 0.04461919888854027, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4618060290813446, + "rewards/pad": 0.0, + "step": 616 + }, + { + "completion_length": 94.0, + "epoch": 0.19662205226258764, + "grad_norm": 71.64122772216797, + "kl": 0.1416015625, + "learning_rate": 8.033779477374123e-07, + "loss": 0.0056, + "reward": 1.4444448947906494, + "reward_std": 0.09489044547080994, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4444449543952942, + "step": 617 + }, + { + "completion_length": 93.359375, + "epoch": 0.19694072657743786, + "grad_norm": 222.44659423828125, + "kl": 0.1396484375, + "learning_rate": 8.03059273422562e-07, + "loss": 0.0056, + "reward": 1.4109399318695068, + "reward_std": 0.03379690647125244, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.41093993186950684, + "rewards/pad": 0.0, + "step": 618 + }, + { + "completion_length": 70.109375, + "epoch": 0.19725940089228808, + "grad_norm": 64.34793853759766, + "kl": 0.1552734375, + "learning_rate": 8.027405991077118e-07, + "loss": 0.0062, + "reward": 1.6567516326904297, + "reward_std": 0.122499480843544, + "rewards/pad": 0.078125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5786266326904297, + "step": 619 + }, + { + "completion_length": 93.859375, + "epoch": 0.1975780752071383, + "grad_norm": 22.94757652282715, + "kl": 0.1728515625, + "learning_rate": 8.024219247928616e-07, + "loss": 0.0069, + "reward": 1.5621228218078613, + "reward_std": 0.08263997733592987, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5621228218078613, + "rewards/pad": 0.0, + "step": 620 + }, + { + "completion_length": 119.5625, + "epoch": 0.19789674952198852, + "grad_norm": 28.08785057067871, + "kl": 0.09228515625, + "learning_rate": 8.021032504780114e-07, + "loss": 0.0037, + "reward": 1.6540088653564453, + "reward_std": 0.05631333217024803, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5290088653564453, + "rewards/pad": 0.125, + "step": 621 + }, + { + "completion_length": 96.734375, + "epoch": 0.19821542383683874, + "grad_norm": 10.783617973327637, + "kl": 0.10888671875, + "learning_rate": 8.017845761631612e-07, + "loss": 0.0044, + "reward": 1.660102367401123, + "reward_std": 0.11361619085073471, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 0.984375, + "rewards/iou_glue_reward": 0.5507273077964783, + "step": 622 + }, + { + "completion_length": 173.765625, + "epoch": 0.198534098151689, + "grad_norm": 33.31416702270508, + "kl": 0.07421875, + "learning_rate": 8.01465901848311e-07, + "loss": 0.003, + "reward": 1.628753900527954, + "reward_std": 0.16182458400726318, + "rewards/pad": 0.234375, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.41000401973724365, + "step": 623 + }, + { + "completion_length": 144.65625, + "epoch": 0.1988527724665392, + "grad_norm": 9.223255157470703, + "kl": 0.287109375, + "learning_rate": 8.011472275334608e-07, + "loss": 0.0115, + "reward": 1.6554591655731201, + "reward_std": 0.060511715710163116, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5304592251777649, + "step": 624 + }, + { + "completion_length": 96.53125, + "epoch": 0.19917144678138943, + "grad_norm": 21.87114906311035, + "kl": 0.11181640625, + "learning_rate": 8.008285532186106e-07, + "loss": 0.0045, + "reward": 1.5894794464111328, + "reward_std": 0.05112447962164879, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5894794464111328, + "step": 625 + }, + { + "completion_length": 71.28125, + "epoch": 0.19949012109623965, + "grad_norm": 19.726043701171875, + "kl": 0.13671875, + "learning_rate": 8.005098789037603e-07, + "loss": 0.0055, + "reward": 1.563927412033081, + "reward_std": 0.06233830377459526, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4389273524284363, + "step": 626 + }, + { + "completion_length": 17.96875, + "epoch": 0.19980879541108987, + "grad_norm": 35.064876556396484, + "kl": 0.41015625, + "learning_rate": 8.001912045889101e-07, + "loss": 0.0165, + "reward": 1.6311964988708496, + "reward_std": 0.12043973058462143, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6311965584754944, + "rewards/pad": 0.0, + "step": 627 + }, + { + "completion_length": 95.609375, + "epoch": 0.2001274697259401, + "grad_norm": 36.62488555908203, + "kl": 0.259765625, + "learning_rate": 7.998725302740599e-07, + "loss": 0.0104, + "reward": 1.4664359092712402, + "reward_std": 0.19014671444892883, + "rewards/answer_reward": 0.03125, + "rewards/format_reward_gqa": 0.984375, + "rewards/iou_glue_reward": 0.450810968875885, + "step": 628 + }, + { + "completion_length": 45.25, + "epoch": 0.2004461440407903, + "grad_norm": 18.387773513793945, + "kl": 0.1767578125, + "learning_rate": 7.995538559592097e-07, + "loss": 0.0071, + "reward": 1.5831278562545776, + "reward_std": 0.07549235969781876, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.45812782645225525, + "rewards/pad": 0.125, + "step": 629 + }, + { + "completion_length": 73.421875, + "epoch": 0.20076481835564053, + "grad_norm": 35.43833541870117, + "kl": 0.16015625, + "learning_rate": 7.992351816443594e-07, + "loss": 0.0064, + "reward": 1.4841811656951904, + "reward_std": 0.12676116824150085, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3591812252998352, + "rewards/pad": 0.125, + "step": 630 + }, + { + "completion_length": 95.703125, + "epoch": 0.20108349267049075, + "grad_norm": 54.55873107910156, + "kl": 0.1337890625, + "learning_rate": 7.989165073295092e-07, + "loss": 0.0053, + "reward": 1.2906908988952637, + "reward_std": 0.05583186075091362, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.2906908690929413, + "step": 631 + }, + { + "completion_length": 97.78125, + "epoch": 0.20140216698534097, + "grad_norm": 25.21178436279297, + "kl": 0.15234375, + "learning_rate": 7.98597833014659e-07, + "loss": 0.0061, + "reward": 1.444516897201538, + "reward_std": 0.0909559354186058, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.44451695680618286, + "step": 632 + }, + { + "completion_length": 96.546875, + "epoch": 0.20172084130019122, + "grad_norm": 34.26476287841797, + "kl": 0.224609375, + "learning_rate": 7.982791586998088e-07, + "loss": 0.009, + "reward": 1.4543168544769287, + "reward_std": 0.08605051785707474, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4543168544769287, + "rewards/pad": 0.0, + "step": 633 + }, + { + "completion_length": 149.78125, + "epoch": 0.20203951561504144, + "grad_norm": 44.93718338012695, + "kl": 0.10205078125, + "learning_rate": 7.979604843849585e-07, + "loss": 0.0041, + "reward": 1.506622076034546, + "reward_std": 0.04655434936285019, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5066221356391907, + "rewards/pad": 0.0, + "step": 634 + }, + { + "completion_length": 45.21875, + "epoch": 0.20235818992989166, + "grad_norm": 60.32438278198242, + "kl": 0.1396484375, + "learning_rate": 7.976418100701083e-07, + "loss": 0.0056, + "reward": 1.6560125350952148, + "reward_std": 0.11041535437107086, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.40601247549057007, + "rewards/pad": 0.25, + "step": 635 + }, + { + "completion_length": 148.0, + "epoch": 0.20267686424474188, + "grad_norm": 13.036028861999512, + "kl": 0.11328125, + "learning_rate": 7.973231357552581e-07, + "loss": 0.0045, + "reward": 1.3512401580810547, + "reward_std": 0.03171146661043167, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3512400686740875, + "rewards/pad": 0.0, + "step": 636 + }, + { + "completion_length": 70.421875, + "epoch": 0.2029955385595921, + "grad_norm": 24.219802856445312, + "kl": 0.255859375, + "learning_rate": 7.970044614404079e-07, + "loss": 0.0102, + "reward": 1.3509703874588013, + "reward_std": 0.1400013566017151, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3509703278541565, + "rewards/pad": 0.0, + "step": 637 + }, + { + "completion_length": 123.578125, + "epoch": 0.20331421287444232, + "grad_norm": 17.177087783813477, + "kl": 0.1171875, + "learning_rate": 7.966857871255576e-07, + "loss": 0.0047, + "reward": 1.5644482374191284, + "reward_std": 0.039421312510967255, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4394482374191284, + "step": 638 + }, + { + "completion_length": 97.078125, + "epoch": 0.20363288718929254, + "grad_norm": 36.70372009277344, + "kl": 0.189453125, + "learning_rate": 7.963671128107074e-07, + "loss": 0.0076, + "reward": 1.605509638786316, + "reward_std": 0.09545377641916275, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6055095195770264, + "rewards/pad": 0.0, + "step": 639 + }, + { + "completion_length": 71.453125, + "epoch": 0.20395156150414276, + "grad_norm": 17.04865837097168, + "kl": 0.26171875, + "learning_rate": 7.960484384958573e-07, + "loss": 0.0105, + "reward": 1.6390584707260132, + "reward_std": 0.14676693081855774, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5140584111213684, + "rewards/pad": 0.125, + "step": 640 + }, + { + "completion_length": 70.640625, + "epoch": 0.20427023581899298, + "grad_norm": 47.15485382080078, + "kl": 0.1767578125, + "learning_rate": 7.957297641810071e-07, + "loss": 0.0071, + "reward": 1.4928598403930664, + "reward_std": 0.08098393678665161, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4928598403930664, + "rewards/pad": 0.0, + "step": 641 + }, + { + "completion_length": 123.046875, + "epoch": 0.2045889101338432, + "grad_norm": 24.090925216674805, + "kl": 0.10693359375, + "learning_rate": 7.954110898661568e-07, + "loss": 0.0043, + "reward": 1.5162612199783325, + "reward_std": 0.027123302221298218, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5162612199783325, + "rewards/pad": 0.0, + "step": 642 + }, + { + "completion_length": 40.46875, + "epoch": 0.20490758444869345, + "grad_norm": 57.921077728271484, + "kl": 0.19921875, + "learning_rate": 7.950924155513066e-07, + "loss": 0.008, + "reward": 1.4682936668395996, + "reward_std": 0.11122076958417892, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.34329357743263245, + "rewards/pad": 0.125, + "step": 643 + }, + { + "completion_length": 149.890625, + "epoch": 0.20522625876354367, + "grad_norm": 100.49211883544922, + "kl": 0.1708984375, + "learning_rate": 7.947737412364564e-07, + "loss": 0.0068, + "reward": 1.3794997930526733, + "reward_std": 0.10549529641866684, + "rewards/answer_reward": 0.015625, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.3638748228549957, + "step": 644 + }, + { + "completion_length": 94.984375, + "epoch": 0.2055449330783939, + "grad_norm": 91.87433624267578, + "kl": 0.2138671875, + "learning_rate": 7.944550669216062e-07, + "loss": 0.0086, + "reward": 1.4812575578689575, + "reward_std": 0.09469930827617645, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3562575578689575, + "rewards/pad": 0.125, + "step": 645 + }, + { + "completion_length": 93.125, + "epoch": 0.2058636073932441, + "grad_norm": 17.375213623046875, + "kl": 0.1708984375, + "learning_rate": 7.941363926067559e-07, + "loss": 0.0068, + "reward": 1.3135367631912231, + "reward_std": 0.04011062905192375, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.31353676319122314, + "rewards/pad": 0.0, + "step": 646 + }, + { + "completion_length": 96.40625, + "epoch": 0.20618228170809433, + "grad_norm": 16.635448455810547, + "kl": 0.162109375, + "learning_rate": 7.938177182919057e-07, + "loss": 0.0065, + "reward": 1.6755173206329346, + "reward_std": 0.14716209471225739, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5505174398422241, + "rewards/pad": 0.125, + "step": 647 + }, + { + "completion_length": 148.796875, + "epoch": 0.20650095602294455, + "grad_norm": 187.33123779296875, + "kl": 0.07861328125, + "learning_rate": 7.934990439770554e-07, + "loss": 0.0031, + "reward": 1.4606094360351562, + "reward_std": 0.09565722942352295, + "rewards/pad": 0.0625, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.39810943603515625, + "step": 648 + }, + { + "completion_length": 73.546875, + "epoch": 0.20681963033779477, + "grad_norm": 36.188865661621094, + "kl": 0.126953125, + "learning_rate": 7.931803696622051e-07, + "loss": 0.0051, + "reward": 1.581721305847168, + "reward_std": 0.05162818729877472, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.45672133564949036, + "rewards/pad": 0.125, + "step": 649 + }, + { + "completion_length": 96.640625, + "epoch": 0.207138304652645, + "grad_norm": 46.26791000366211, + "kl": 0.162109375, + "learning_rate": 7.928616953473549e-07, + "loss": 0.0065, + "reward": 1.7344567775726318, + "reward_std": 0.10052649676799774, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6094566583633423, + "rewards/pad": 0.125, + "step": 650 + }, + { + "completion_length": 68.734375, + "epoch": 0.2074569789674952, + "grad_norm": 52.092620849609375, + "kl": 0.474609375, + "learning_rate": 7.925430210325047e-07, + "loss": 0.019, + "reward": 1.5348825454711914, + "reward_std": 0.11189387738704681, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.40988245606422424, + "rewards/pad": 0.125, + "step": 651 + }, + { + "completion_length": 96.078125, + "epoch": 0.20777565328234543, + "grad_norm": 100.03245544433594, + "kl": 0.1376953125, + "learning_rate": 7.922243467176545e-07, + "loss": 0.0055, + "reward": 1.4961674213409424, + "reward_std": 0.09468173235654831, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.5117923617362976, + "rewards/pad": 0.0, + "step": 652 + }, + { + "completion_length": 97.484375, + "epoch": 0.20809432759719568, + "grad_norm": 35.96536636352539, + "kl": 0.1435546875, + "learning_rate": 7.919056724028042e-07, + "loss": 0.0057, + "reward": 1.52105712890625, + "reward_std": 0.045935630798339844, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3960571587085724, + "rewards/pad": 0.125, + "step": 653 + }, + { + "completion_length": 97.75, + "epoch": 0.2084130019120459, + "grad_norm": 30.760404586791992, + "kl": 0.1884765625, + "learning_rate": 7.91586998087954e-07, + "loss": 0.0075, + "reward": 1.5083422660827637, + "reward_std": 0.11229605227708817, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5083421468734741, + "rewards/pad": 0.0, + "step": 654 + }, + { + "completion_length": 146.828125, + "epoch": 0.20873167622689612, + "grad_norm": 251.43666076660156, + "kl": 0.10595703125, + "learning_rate": 7.912683237731038e-07, + "loss": 0.0042, + "reward": 1.4129326343536377, + "reward_std": 0.04007997736334801, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4129326045513153, + "step": 655 + }, + { + "completion_length": 71.875, + "epoch": 0.20905035054174634, + "grad_norm": 22.842525482177734, + "kl": 0.2392578125, + "learning_rate": 7.909496494582536e-07, + "loss": 0.0096, + "reward": 1.6012282371520996, + "reward_std": 0.08485373854637146, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5856032371520996, + "rewards/pad": 0.015625, + "step": 656 + }, + { + "completion_length": 124.265625, + "epoch": 0.20936902485659656, + "grad_norm": 25.405227661132812, + "kl": 0.2001953125, + "learning_rate": 7.906309751434033e-07, + "loss": 0.008, + "reward": 1.6624181270599365, + "reward_std": 0.08725246042013168, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4124181568622589, + "rewards/pad": 0.25, + "step": 657 + }, + { + "completion_length": 71.84375, + "epoch": 0.20968769917144678, + "grad_norm": 47.46111297607422, + "kl": 0.189453125, + "learning_rate": 7.903123008285531e-07, + "loss": 0.0076, + "reward": 1.605202555656433, + "reward_std": 0.10207031667232513, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.35520249605178833, + "rewards/pad": 0.25, + "step": 658 + }, + { + "completion_length": 125.53125, + "epoch": 0.210006373486297, + "grad_norm": 24.989748001098633, + "kl": 0.12890625, + "learning_rate": 7.89993626513703e-07, + "loss": 0.0052, + "reward": 1.513319492340088, + "reward_std": 0.053028907626867294, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5133194923400879, + "rewards/pad": 0.0, + "step": 659 + }, + { + "completion_length": 98.125, + "epoch": 0.21032504780114722, + "grad_norm": 18.915529251098633, + "kl": 0.2080078125, + "learning_rate": 7.896749521988528e-07, + "loss": 0.0083, + "reward": 1.4596118927001953, + "reward_std": 0.09722685813903809, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4596119821071625, + "rewards/pad": 0.0, + "step": 660 + }, + { + "completion_length": 97.546875, + "epoch": 0.21064372211599744, + "grad_norm": 114.42607879638672, + "kl": 0.12060546875, + "learning_rate": 7.893562778840025e-07, + "loss": 0.0048, + "reward": 1.5191466808319092, + "reward_std": 0.09039562195539474, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.34727177023887634, + "rewards/pad": 0.171875, + "step": 661 + }, + { + "completion_length": 126.953125, + "epoch": 0.2109623964308477, + "grad_norm": 19.043169021606445, + "kl": 0.09619140625, + "learning_rate": 7.890376035691523e-07, + "loss": 0.0038, + "reward": 1.7435672283172607, + "reward_std": 0.046088241040706635, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4935673475265503, + "rewards/pad": 0.25, + "step": 662 + }, + { + "completion_length": 69.75, + "epoch": 0.2112810707456979, + "grad_norm": 25.736791610717773, + "kl": 0.212890625, + "learning_rate": 7.887189292543021e-07, + "loss": 0.0085, + "reward": 1.5893003940582275, + "reward_std": 0.07170729339122772, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5893005132675171, + "rewards/pad": 0.0, + "step": 663 + }, + { + "completion_length": 97.71875, + "epoch": 0.21159974506054813, + "grad_norm": 66.05033111572266, + "kl": 0.1552734375, + "learning_rate": 7.884002549394519e-07, + "loss": 0.0062, + "reward": 1.508967399597168, + "reward_std": 0.0578170046210289, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5089674592018127, + "rewards/pad": 0.0, + "step": 664 + }, + { + "completion_length": 97.90625, + "epoch": 0.21191841937539835, + "grad_norm": 34.869117736816406, + "kl": 0.10205078125, + "learning_rate": 7.880815806246016e-07, + "loss": 0.0041, + "reward": 1.3681917190551758, + "reward_std": 0.057296961545944214, + "rewards/answer_reward": 0.0, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.3681916892528534, + "step": 665 + }, + { + "completion_length": 123.03125, + "epoch": 0.21223709369024857, + "grad_norm": 25.530683517456055, + "kl": 0.1123046875, + "learning_rate": 7.877629063097514e-07, + "loss": 0.0045, + "reward": 1.5382572412490845, + "reward_std": 0.07736440747976303, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4132571816444397, + "step": 666 + }, + { + "completion_length": 96.1875, + "epoch": 0.2125557680050988, + "grad_norm": 27.343393325805664, + "kl": 0.212890625, + "learning_rate": 7.874442319949012e-07, + "loss": 0.0085, + "reward": 1.3758732080459595, + "reward_std": 0.0433371402323246, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3758732080459595, + "rewards/pad": 0.0, + "step": 667 + }, + { + "completion_length": 126.1875, + "epoch": 0.212874442319949, + "grad_norm": 25.15089988708496, + "kl": 0.08154296875, + "learning_rate": 7.87125557680051e-07, + "loss": 0.0033, + "reward": 1.8917670249938965, + "reward_std": 0.04995495826005936, + "rewards/answer_reward": 0.375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5167669653892517, + "step": 668 + }, + { + "completion_length": 146.21875, + "epoch": 0.21319311663479923, + "grad_norm": 39.95366287231445, + "kl": 0.12109375, + "learning_rate": 7.868068833652007e-07, + "loss": 0.0048, + "reward": 1.496253252029419, + "reward_std": 0.06662138551473618, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.49625325202941895, + "step": 669 + }, + { + "completion_length": 98.21875, + "epoch": 0.21351179094964945, + "grad_norm": 17.594482421875, + "kl": 0.22265625, + "learning_rate": 7.864882090503505e-07, + "loss": 0.0089, + "reward": 1.582715630531311, + "reward_std": 0.06213941425085068, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5827155709266663, + "rewards/pad": 0.0, + "step": 670 + }, + { + "completion_length": 72.796875, + "epoch": 0.21383046526449967, + "grad_norm": 43.37171936035156, + "kl": 0.1728515625, + "learning_rate": 7.861695347355003e-07, + "loss": 0.0069, + "reward": 1.473966121673584, + "reward_std": 0.09620095789432526, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.45834115147590637, + "rewards/pad": 0.015625, + "step": 671 + }, + { + "completion_length": 96.96875, + "epoch": 0.21414913957934992, + "grad_norm": 34.0592155456543, + "kl": 0.1015625, + "learning_rate": 7.858508604206501e-07, + "loss": 0.0041, + "reward": 1.7571685314178467, + "reward_std": 0.0643969252705574, + "rewards/pad": 0.375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3821684420108795, + "step": 672 + }, + { + "completion_length": 123.1875, + "epoch": 0.21446781389420014, + "grad_norm": 137.472412109375, + "kl": 0.138671875, + "learning_rate": 7.855321861057998e-07, + "loss": 0.0056, + "reward": 1.4088833332061768, + "reward_std": 0.07924804091453552, + "rewards/answer_reward": 0.0, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.40888330340385437, + "step": 673 + }, + { + "completion_length": 123.8125, + "epoch": 0.21478648820905036, + "grad_norm": 20.52865982055664, + "kl": 0.11572265625, + "learning_rate": 7.852135117909496e-07, + "loss": 0.0046, + "reward": 1.6505613327026367, + "reward_std": 0.053364917635917664, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5255613923072815, + "step": 674 + }, + { + "completion_length": 98.140625, + "epoch": 0.21510516252390058, + "grad_norm": 25.846698760986328, + "kl": 0.197265625, + "learning_rate": 7.848948374760994e-07, + "loss": 0.0079, + "reward": 1.5987257957458496, + "reward_std": 0.13354355096817017, + "rewards/answer_reward": 0.03125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5674759149551392, + "step": 675 + }, + { + "completion_length": 73.765625, + "epoch": 0.2154238368387508, + "grad_norm": 134.68280029296875, + "kl": 0.1416015625, + "learning_rate": 7.845761631612492e-07, + "loss": 0.0056, + "reward": 1.5867506265640259, + "reward_std": 0.10678602755069733, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3367506265640259, + "rewards/pad": 0.25, + "step": 676 + }, + { + "completion_length": 68.921875, + "epoch": 0.21574251115360102, + "grad_norm": 25.11921501159668, + "kl": 0.2275390625, + "learning_rate": 7.842574888463989e-07, + "loss": 0.0091, + "reward": 1.5466156005859375, + "reward_std": 0.05886758118867874, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5466156005859375, + "step": 677 + }, + { + "completion_length": 122.078125, + "epoch": 0.21606118546845124, + "grad_norm": 41.1529426574707, + "kl": 0.1162109375, + "learning_rate": 7.839388145315488e-07, + "loss": 0.0046, + "reward": 1.5090479850769043, + "reward_std": 0.0427854061126709, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5090479850769043, + "step": 678 + }, + { + "completion_length": 46.921875, + "epoch": 0.21637985978330146, + "grad_norm": 32.37312316894531, + "kl": 0.1474609375, + "learning_rate": 7.836201402166986e-07, + "loss": 0.0059, + "reward": 2.004756450653076, + "reward_std": 0.09021234512329102, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6297565698623657, + "rewards/pad": 0.375, + "step": 679 + }, + { + "completion_length": 126.328125, + "epoch": 0.21669853409815168, + "grad_norm": 71.86237335205078, + "kl": 0.07958984375, + "learning_rate": 7.833014659018483e-07, + "loss": 0.0032, + "reward": 1.5416804552078247, + "reward_std": 0.041763074696063995, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4166804552078247, + "rewards/pad": 0.125, + "step": 680 + }, + { + "completion_length": 121.59375, + "epoch": 0.2170172084130019, + "grad_norm": 16.156023025512695, + "kl": 0.0859375, + "learning_rate": 7.829827915869981e-07, + "loss": 0.0034, + "reward": 1.3597514629364014, + "reward_std": 0.06102152541279793, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.35975152254104614, + "step": 681 + }, + { + "completion_length": 125.03125, + "epoch": 0.21733588272785215, + "grad_norm": 40.038936614990234, + "kl": 0.07666015625, + "learning_rate": 7.826641172721479e-07, + "loss": 0.0031, + "reward": 1.729158639907837, + "reward_std": 0.03963400423526764, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.47915855050086975, + "rewards/pad": 0.25, + "step": 682 + }, + { + "completion_length": 96.21875, + "epoch": 0.21765455704270237, + "grad_norm": 85.6227798461914, + "kl": 0.15234375, + "learning_rate": 7.823454429572977e-07, + "loss": 0.0061, + "reward": 1.4844729900360107, + "reward_std": 0.07393090426921844, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.48447293043136597, + "step": 683 + }, + { + "completion_length": 121.40625, + "epoch": 0.2179732313575526, + "grad_norm": 11.68277359008789, + "kl": 0.0966796875, + "learning_rate": 7.820267686424474e-07, + "loss": 0.0039, + "reward": 1.4158823490142822, + "reward_std": 0.04511553421616554, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4158823788166046, + "rewards/pad": 0.0, + "step": 684 + }, + { + "completion_length": 146.984375, + "epoch": 0.2182919056724028, + "grad_norm": 8.718667984008789, + "kl": 0.046630859375, + "learning_rate": 7.817080943275972e-07, + "loss": 0.0019, + "reward": 1.6158006191253662, + "reward_std": 0.03874984011054039, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3658006191253662, + "step": 685 + }, + { + "completion_length": 98.03125, + "epoch": 0.21861057998725303, + "grad_norm": 12.766626358032227, + "kl": 0.13671875, + "learning_rate": 7.81389420012747e-07, + "loss": 0.0055, + "reward": 1.731637954711914, + "reward_std": 0.05565390735864639, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6066379547119141, + "rewards/pad": 0.125, + "step": 686 + }, + { + "completion_length": 125.109375, + "epoch": 0.21892925430210325, + "grad_norm": 19.363597869873047, + "kl": 0.12353515625, + "learning_rate": 7.810707456978967e-07, + "loss": 0.0049, + "reward": 1.555992841720581, + "reward_std": 0.08267258107662201, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.555992841720581, + "rewards/pad": 0.0, + "step": 687 + }, + { + "completion_length": 98.484375, + "epoch": 0.21924792861695347, + "grad_norm": 29.480148315429688, + "kl": 0.16015625, + "learning_rate": 7.807520713830464e-07, + "loss": 0.0064, + "reward": 1.4770445823669434, + "reward_std": 0.0575791671872139, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.35204464197158813, + "rewards/pad": 0.125, + "step": 688 + }, + { + "completion_length": 150.578125, + "epoch": 0.2195666029318037, + "grad_norm": 33.247291564941406, + "kl": 0.1279296875, + "learning_rate": 7.804333970681962e-07, + "loss": 0.0051, + "reward": 1.4387083053588867, + "reward_std": 0.04876134172081947, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3137083351612091, + "step": 689 + }, + { + "completion_length": 99.796875, + "epoch": 0.2198852772466539, + "grad_norm": 19.67684555053711, + "kl": 0.11181640625, + "learning_rate": 7.80114722753346e-07, + "loss": 0.0045, + "reward": 1.576588749885559, + "reward_std": 0.06405520439147949, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4515886902809143, + "rewards/pad": 0.125, + "step": 690 + }, + { + "completion_length": 70.453125, + "epoch": 0.22020395156150413, + "grad_norm": 17.070262908935547, + "kl": 0.1552734375, + "learning_rate": 7.797960484384958e-07, + "loss": 0.0062, + "reward": 1.4112935066223145, + "reward_std": 0.06292285025119781, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4112934470176697, + "step": 691 + }, + { + "completion_length": 92.796875, + "epoch": 0.22052262587635438, + "grad_norm": 8.107675552368164, + "kl": 0.16015625, + "learning_rate": 7.794773741236455e-07, + "loss": 0.0064, + "reward": 1.4180748462677002, + "reward_std": 0.07736538350582123, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4180748164653778, + "rewards/pad": 0.0, + "step": 692 + }, + { + "completion_length": 98.6875, + "epoch": 0.2208413001912046, + "grad_norm": 38.32047653198242, + "kl": 0.19140625, + "learning_rate": 7.791586998087953e-07, + "loss": 0.0077, + "reward": 1.3982741832733154, + "reward_std": 0.06808879971504211, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3982742428779602, + "rewards/pad": 0.0, + "step": 693 + }, + { + "completion_length": 148.578125, + "epoch": 0.22115997450605482, + "grad_norm": 32.58980941772461, + "kl": 0.115234375, + "learning_rate": 7.788400254939451e-07, + "loss": 0.0046, + "reward": 1.4551475048065186, + "reward_std": 0.0770447701215744, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.45514750480651855, + "step": 694 + }, + { + "completion_length": 149.90625, + "epoch": 0.22147864882090504, + "grad_norm": 21.80135154724121, + "kl": 0.1064453125, + "learning_rate": 7.785213511790949e-07, + "loss": 0.0043, + "reward": 1.4896881580352783, + "reward_std": 0.05898185819387436, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.36468809843063354, + "step": 695 + }, + { + "completion_length": 46.96875, + "epoch": 0.22179732313575526, + "grad_norm": 161.15521240234375, + "kl": 0.1484375, + "learning_rate": 7.782026768642446e-07, + "loss": 0.0059, + "reward": 1.559443712234497, + "reward_std": 0.1841370314359665, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3719435930252075, + "rewards/pad": 0.1875, + "step": 696 + }, + { + "completion_length": 97.328125, + "epoch": 0.22211599745060548, + "grad_norm": 43.470455169677734, + "kl": 0.11376953125, + "learning_rate": 7.778840025493945e-07, + "loss": 0.0045, + "reward": 1.5548875331878662, + "reward_std": 0.13499760627746582, + "rewards/answer_reward": 0.015625, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5392624139785767, + "step": 697 + }, + { + "completion_length": 149.1875, + "epoch": 0.2224346717654557, + "grad_norm": 46.18128967285156, + "kl": 0.078125, + "learning_rate": 7.775653282345443e-07, + "loss": 0.0031, + "reward": 1.4198707342147827, + "reward_std": 0.062181681394577026, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4198707640171051, + "step": 698 + }, + { + "completion_length": 97.953125, + "epoch": 0.22275334608030592, + "grad_norm": 26.341793060302734, + "kl": 0.12451171875, + "learning_rate": 7.772466539196941e-07, + "loss": 0.005, + "reward": 1.64949369430542, + "reward_std": 0.1283523142337799, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.5401187539100647, + "step": 699 + }, + { + "completion_length": 146.609375, + "epoch": 0.22307202039515614, + "grad_norm": 10.743393898010254, + "kl": 0.05810546875, + "learning_rate": 7.769279796048438e-07, + "loss": 0.0023, + "reward": 1.4474786520004272, + "reward_std": 0.03060361184179783, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.44747862219810486, + "rewards/pad": 0.0, + "step": 700 + }, + { + "completion_length": 123.90625, + "epoch": 0.22339069471000636, + "grad_norm": 22.797927856445312, + "kl": 0.357421875, + "learning_rate": 7.766093052899936e-07, + "loss": 0.0143, + "reward": 1.7013278007507324, + "reward_std": 0.08491910994052887, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5763278603553772, + "rewards/pad": 0.125, + "step": 701 + }, + { + "completion_length": 71.453125, + "epoch": 0.2237093690248566, + "grad_norm": 17.504398345947266, + "kl": 0.27734375, + "learning_rate": 7.762906309751434e-07, + "loss": 0.0111, + "reward": 1.6171817779541016, + "reward_std": 0.11721836030483246, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6171817779541016, + "rewards/pad": 0.0, + "step": 702 + }, + { + "completion_length": 46.109375, + "epoch": 0.22402804333970683, + "grad_norm": 40.05861282348633, + "kl": 0.1328125, + "learning_rate": 7.759719566602932e-07, + "loss": 0.0053, + "reward": 1.6811028718948364, + "reward_std": 0.08720278739929199, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4311029613018036, + "step": 703 + }, + { + "completion_length": 151.890625, + "epoch": 0.22434671765455705, + "grad_norm": 27.30905532836914, + "kl": 0.1025390625, + "learning_rate": 7.756532823454429e-07, + "loss": 0.0041, + "reward": 1.602893352508545, + "reward_std": 0.12148568034172058, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.4935183823108673, + "rewards/pad": 0.125, + "step": 704 + }, + { + "completion_length": 121.765625, + "epoch": 0.22466539196940727, + "grad_norm": 20.993581771850586, + "kl": 0.21484375, + "learning_rate": 7.753346080305927e-07, + "loss": 0.0086, + "reward": 1.5430442094802856, + "reward_std": 0.09577062726020813, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5430442690849304, + "rewards/pad": 0.0, + "step": 705 + }, + { + "completion_length": 21.5625, + "epoch": 0.2249840662842575, + "grad_norm": 50.83089828491211, + "kl": 0.15234375, + "learning_rate": 7.750159337157425e-07, + "loss": 0.0061, + "reward": 1.6595174074172974, + "reward_std": 0.07145840674638748, + "rewards/answer_reward": 0.375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.2845174968242645, + "step": 706 + }, + { + "completion_length": 98.28125, + "epoch": 0.2253027405991077, + "grad_norm": 35.206878662109375, + "kl": 0.115234375, + "learning_rate": 7.746972594008923e-07, + "loss": 0.0046, + "reward": 1.5010448694229126, + "reward_std": 0.11892069876194, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.2822949290275574, + "rewards/pad": 0.21875, + "step": 707 + }, + { + "completion_length": 96.734375, + "epoch": 0.22562141491395793, + "grad_norm": 97.78717803955078, + "kl": 0.1435546875, + "learning_rate": 7.74378585086042e-07, + "loss": 0.0057, + "reward": 1.5857737064361572, + "reward_std": 0.13069474697113037, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.460773766040802, + "rewards/pad": 0.125, + "step": 708 + }, + { + "completion_length": 45.015625, + "epoch": 0.22594008922880815, + "grad_norm": 37.87092208862305, + "kl": 0.1396484375, + "learning_rate": 7.740599107711918e-07, + "loss": 0.0056, + "reward": 1.6845557689666748, + "reward_std": 0.0830632746219635, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5595556497573853, + "step": 709 + }, + { + "completion_length": 73.625, + "epoch": 0.22625876354365837, + "grad_norm": 21.10732650756836, + "kl": 0.142578125, + "learning_rate": 7.737412364563416e-07, + "loss": 0.0057, + "reward": 1.508007526397705, + "reward_std": 0.15659797191619873, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.5236325263977051, + "step": 710 + }, + { + "completion_length": 46.296875, + "epoch": 0.22657743785850862, + "grad_norm": 91.6818618774414, + "kl": 0.1669921875, + "learning_rate": 7.734225621414913e-07, + "loss": 0.0067, + "reward": 1.6029472351074219, + "reward_std": 0.11285988241434097, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.47794726490974426, + "step": 711 + }, + { + "completion_length": 123.28125, + "epoch": 0.22689611217335884, + "grad_norm": 85.90867614746094, + "kl": 0.263671875, + "learning_rate": 7.731038878266411e-07, + "loss": 0.0106, + "reward": 1.525963544845581, + "reward_std": 0.09025786817073822, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5259636640548706, + "step": 712 + }, + { + "completion_length": 123.890625, + "epoch": 0.22721478648820906, + "grad_norm": 61.534950256347656, + "kl": 0.0791015625, + "learning_rate": 7.727852135117909e-07, + "loss": 0.0032, + "reward": 1.6034601926803589, + "reward_std": 0.07537369430065155, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4784601330757141, + "rewards/pad": 0.125, + "step": 713 + }, + { + "completion_length": 98.46875, + "epoch": 0.22753346080305928, + "grad_norm": 45.14211654663086, + "kl": 0.1328125, + "learning_rate": 7.724665391969407e-07, + "loss": 0.0053, + "reward": 1.4492158889770508, + "reward_std": 0.14678949117660522, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4179658591747284, + "rewards/pad": 0.03125, + "step": 714 + }, + { + "completion_length": 72.203125, + "epoch": 0.2278521351179095, + "grad_norm": 30.922330856323242, + "kl": 0.15234375, + "learning_rate": 7.721478648820904e-07, + "loss": 0.0061, + "reward": 1.5465211868286133, + "reward_std": 0.08825637400150299, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5465211868286133, + "rewards/pad": 0.0, + "step": 715 + }, + { + "completion_length": 175.65625, + "epoch": 0.22817080943275972, + "grad_norm": 17.37681770324707, + "kl": 0.04443359375, + "learning_rate": 7.718291905672403e-07, + "loss": 0.0018, + "reward": 1.5493981838226318, + "reward_std": 0.06132122129201889, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4243982434272766, + "step": 716 + }, + { + "completion_length": 124.8125, + "epoch": 0.22848948374760994, + "grad_norm": 31.47076988220215, + "kl": 0.09521484375, + "learning_rate": 7.715105162523901e-07, + "loss": 0.0038, + "reward": 1.6855530738830566, + "reward_std": 0.05492226034402847, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5605530738830566, + "rewards/pad": 0.125, + "step": 717 + }, + { + "completion_length": 71.03125, + "epoch": 0.22880815806246016, + "grad_norm": 23.651077270507812, + "kl": 0.21484375, + "learning_rate": 7.711918419375399e-07, + "loss": 0.0086, + "reward": 1.5401701927185059, + "reward_std": 0.06382852792739868, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5401701331138611, + "step": 718 + }, + { + "completion_length": 70.734375, + "epoch": 0.22912683237731038, + "grad_norm": 27.53903579711914, + "kl": 0.22265625, + "learning_rate": 7.708731676226896e-07, + "loss": 0.0089, + "reward": 1.4649901390075684, + "reward_std": 0.13506777584552765, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4649902284145355, + "rewards/pad": 0.0, + "step": 719 + }, + { + "completion_length": 72.0625, + "epoch": 0.2294455066921606, + "grad_norm": 31.68392562866211, + "kl": 0.126953125, + "learning_rate": 7.705544933078394e-07, + "loss": 0.0051, + "reward": 1.4837205410003662, + "reward_std": 0.07027556002140045, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.358720600605011, + "step": 720 + }, + { + "completion_length": 94.421875, + "epoch": 0.22976418100701085, + "grad_norm": 36.224647521972656, + "kl": 0.130859375, + "learning_rate": 7.702358189929892e-07, + "loss": 0.0052, + "reward": 1.4542423486709595, + "reward_std": 0.08611422777175903, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4542423486709595, + "rewards/pad": 0.0, + "step": 721 + }, + { + "completion_length": 118.578125, + "epoch": 0.23008285532186107, + "grad_norm": 27.81844139099121, + "kl": 0.11572265625, + "learning_rate": 7.69917144678139e-07, + "loss": 0.0046, + "reward": 1.655103087425232, + "reward_std": 0.0717756524682045, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5301031470298767, + "step": 722 + }, + { + "completion_length": 100.34375, + "epoch": 0.2304015296367113, + "grad_norm": 17.767757415771484, + "kl": 0.11083984375, + "learning_rate": 7.695984703632887e-07, + "loss": 0.0044, + "reward": 1.4792160987854004, + "reward_std": 0.06914187967777252, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.35421618819236755, + "rewards/pad": 0.125, + "step": 723 + }, + { + "completion_length": 123.5625, + "epoch": 0.2307202039515615, + "grad_norm": 48.519046783447266, + "kl": 0.1357421875, + "learning_rate": 7.692797960484385e-07, + "loss": 0.0054, + "reward": 1.6084320545196533, + "reward_std": 0.13084352016448975, + "rewards/answer_reward": 0.0625, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5459321737289429, + "step": 724 + }, + { + "completion_length": 151.90625, + "epoch": 0.23103887826641173, + "grad_norm": 17.062904357910156, + "kl": 0.0634765625, + "learning_rate": 7.689611217335883e-07, + "loss": 0.0025, + "reward": 1.7455984354019165, + "reward_std": 0.06851904094219208, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4955984652042389, + "step": 725 + }, + { + "completion_length": 122.59375, + "epoch": 0.23135755258126195, + "grad_norm": 86.39067077636719, + "kl": 0.1064453125, + "learning_rate": 7.68642447418738e-07, + "loss": 0.0043, + "reward": 1.4053685665130615, + "reward_std": 0.08708634972572327, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4053686261177063, + "step": 726 + }, + { + "completion_length": 70.765625, + "epoch": 0.23167622689611217, + "grad_norm": 253.50587463378906, + "kl": 0.25, + "learning_rate": 7.683237731038877e-07, + "loss": 0.01, + "reward": 1.5726487636566162, + "reward_std": 0.11331886053085327, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.572648823261261, + "rewards/pad": 0.0, + "step": 727 + }, + { + "completion_length": 70.640625, + "epoch": 0.2319949012109624, + "grad_norm": 20.981658935546875, + "kl": 0.1337890625, + "learning_rate": 7.680050987890375e-07, + "loss": 0.0053, + "reward": 1.7513389587402344, + "reward_std": 0.0925459936261177, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6263389587402344, + "rewards/pad": 0.125, + "step": 728 + }, + { + "completion_length": 149.046875, + "epoch": 0.2323135755258126, + "grad_norm": 19.781028747558594, + "kl": 0.068359375, + "learning_rate": 7.676864244741873e-07, + "loss": 0.0027, + "reward": 1.5592389106750488, + "reward_std": 0.038696493953466415, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4342389404773712, + "step": 729 + }, + { + "completion_length": 69.6875, + "epoch": 0.23263224984066283, + "grad_norm": 32.08163833618164, + "kl": 0.09619140625, + "learning_rate": 7.673677501593371e-07, + "loss": 0.0038, + "reward": 1.599071979522705, + "reward_std": 0.06046079471707344, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5990719795227051, + "rewards/pad": 0.0, + "step": 730 + }, + { + "completion_length": 71.109375, + "epoch": 0.23295092415551308, + "grad_norm": 55.546295166015625, + "kl": 0.16015625, + "learning_rate": 7.670490758444868e-07, + "loss": 0.0064, + "reward": 1.6382114887237549, + "reward_std": 0.12515056133270264, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5132115483283997, + "step": 731 + }, + { + "completion_length": 45.84375, + "epoch": 0.2332695984703633, + "grad_norm": 19.57425880432129, + "kl": 0.205078125, + "learning_rate": 7.667304015296366e-07, + "loss": 0.0082, + "reward": 1.5063813924789429, + "reward_std": 0.07701030373573303, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5063812732696533, + "rewards/pad": 0.0, + "step": 732 + }, + { + "completion_length": 176.40625, + "epoch": 0.23358827278521352, + "grad_norm": 15.526483535766602, + "kl": 0.06591796875, + "learning_rate": 7.664117272147864e-07, + "loss": 0.0026, + "reward": 1.5167168378829956, + "reward_std": 0.06336955726146698, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5167168974876404, + "rewards/pad": 0.0, + "step": 733 + }, + { + "completion_length": 122.890625, + "epoch": 0.23390694710006374, + "grad_norm": 20.98141098022461, + "kl": 0.08447265625, + "learning_rate": 7.660930528999362e-07, + "loss": 0.0034, + "reward": 1.6783380508422852, + "reward_std": 0.07113255560398102, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5533379316329956, + "rewards/pad": 0.125, + "step": 734 + }, + { + "completion_length": 98.4375, + "epoch": 0.23422562141491396, + "grad_norm": 34.61638259887695, + "kl": 0.1943359375, + "learning_rate": 7.65774378585086e-07, + "loss": 0.0077, + "reward": 1.6574945449829102, + "reward_std": 0.08876973390579224, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.40749454498291016, + "rewards/pad": 0.25, + "step": 735 + }, + { + "completion_length": 71.28125, + "epoch": 0.23454429572976418, + "grad_norm": 119.02059936523438, + "kl": 0.1787109375, + "learning_rate": 7.654557042702358e-07, + "loss": 0.0072, + "reward": 1.688605785369873, + "reward_std": 0.09231105446815491, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.563605785369873, + "step": 736 + }, + { + "completion_length": 101.03125, + "epoch": 0.2348629700446144, + "grad_norm": 47.49235153198242, + "kl": 0.107421875, + "learning_rate": 7.651370299553856e-07, + "loss": 0.0043, + "reward": 1.6657613515853882, + "reward_std": 0.058830223977565765, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.540761411190033, + "step": 737 + }, + { + "completion_length": 20.640625, + "epoch": 0.23518164435946462, + "grad_norm": 181.3784637451172, + "kl": 0.2109375, + "learning_rate": 7.648183556405353e-07, + "loss": 0.0084, + "reward": 1.7108551263809204, + "reward_std": 0.0872071236371994, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3358551263809204, + "rewards/pad": 0.375, + "step": 738 + }, + { + "completion_length": 45.609375, + "epoch": 0.23550031867431484, + "grad_norm": 47.21954345703125, + "kl": 0.1396484375, + "learning_rate": 7.644996813256851e-07, + "loss": 0.0056, + "reward": 1.4048912525177002, + "reward_std": 0.09731826186180115, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.40489131212234497, + "rewards/pad": 0.0, + "step": 739 + }, + { + "completion_length": 149.640625, + "epoch": 0.23581899298916506, + "grad_norm": 21.183887481689453, + "kl": 0.09521484375, + "learning_rate": 7.641810070108349e-07, + "loss": 0.0038, + "reward": 1.4415016174316406, + "reward_std": 0.09589053690433502, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.44150158762931824, + "step": 740 + }, + { + "completion_length": 97.921875, + "epoch": 0.2361376673040153, + "grad_norm": 84.9200439453125, + "kl": 0.11669921875, + "learning_rate": 7.638623326959847e-07, + "loss": 0.0047, + "reward": 1.4293819665908813, + "reward_std": 0.07976148277521133, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.42938196659088135, + "step": 741 + }, + { + "completion_length": 72.359375, + "epoch": 0.23645634161886553, + "grad_norm": 28.53392791748047, + "kl": 0.1396484375, + "learning_rate": 7.635436583811344e-07, + "loss": 0.0056, + "reward": 1.4849293231964111, + "reward_std": 0.05757514387369156, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3599294126033783, + "rewards/pad": 0.125, + "step": 742 + }, + { + "completion_length": 46.703125, + "epoch": 0.23677501593371575, + "grad_norm": 44.556705474853516, + "kl": 0.1376953125, + "learning_rate": 7.632249840662842e-07, + "loss": 0.0055, + "reward": 1.5273563861846924, + "reward_std": 0.08845867216587067, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.40235644578933716, + "rewards/pad": 0.125, + "step": 743 + }, + { + "completion_length": 123.4375, + "epoch": 0.23709369024856597, + "grad_norm": 45.98194122314453, + "kl": 0.10009765625, + "learning_rate": 7.62906309751434e-07, + "loss": 0.004, + "reward": 1.3977434635162354, + "reward_std": 0.043766941875219345, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3977435231208801, + "step": 744 + }, + { + "completion_length": 72.25, + "epoch": 0.2374123645634162, + "grad_norm": 16.99376106262207, + "kl": 0.2060546875, + "learning_rate": 7.625876354365838e-07, + "loss": 0.0082, + "reward": 1.5830817222595215, + "reward_std": 0.03701595962047577, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.45808184146881104, + "step": 745 + }, + { + "completion_length": 121.546875, + "epoch": 0.2377310388782664, + "grad_norm": 47.60268020629883, + "kl": 0.09716796875, + "learning_rate": 7.622689611217335e-07, + "loss": 0.0039, + "reward": 1.6375110149383545, + "reward_std": 0.07031754404306412, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5125110745429993, + "step": 746 + }, + { + "completion_length": 151.84375, + "epoch": 0.23804971319311663, + "grad_norm": 24.645835876464844, + "kl": 0.03662109375, + "learning_rate": 7.619502868068833e-07, + "loss": 0.0015, + "reward": 1.6020203828811646, + "reward_std": 0.03844858705997467, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4770203232765198, + "step": 747 + }, + { + "completion_length": 74.140625, + "epoch": 0.23836838750796685, + "grad_norm": 51.959228515625, + "kl": 0.1005859375, + "learning_rate": 7.616316124920331e-07, + "loss": 0.004, + "reward": 1.730022668838501, + "reward_std": 0.1415863335132599, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.43314772844314575, + "rewards/pad": 0.296875, + "step": 748 + }, + { + "completion_length": 71.375, + "epoch": 0.23868706182281707, + "grad_norm": 108.53982543945312, + "kl": 0.267578125, + "learning_rate": 7.613129381771829e-07, + "loss": 0.0107, + "reward": 1.4023091793060303, + "reward_std": 0.09092642366886139, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4023091197013855, + "rewards/pad": 0.0, + "step": 749 + }, + { + "completion_length": 173.0625, + "epoch": 0.2390057361376673, + "grad_norm": 9.18797779083252, + "kl": 0.06640625, + "learning_rate": 7.609942638623326e-07, + "loss": 0.0027, + "reward": 1.4714884757995605, + "reward_std": 0.09247317910194397, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.4871135354042053, + "rewards/pad": 0.0, + "step": 750 + }, + { + "completion_length": 123.609375, + "epoch": 0.23932441045251754, + "grad_norm": 15.563135147094727, + "kl": 0.11474609375, + "learning_rate": 7.606755895474824e-07, + "loss": 0.0046, + "reward": 1.6107486486434937, + "reward_std": 0.0644153356552124, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4857487380504608, + "rewards/pad": 0.125, + "step": 751 + }, + { + "completion_length": 122.96875, + "epoch": 0.23964308476736776, + "grad_norm": 40.96657943725586, + "kl": 0.07177734375, + "learning_rate": 7.603569152326322e-07, + "loss": 0.0029, + "reward": 1.460272192955017, + "reward_std": 0.05413992702960968, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4602721631526947, + "rewards/pad": 0.0, + "step": 752 + }, + { + "completion_length": 72.1875, + "epoch": 0.23996175908221798, + "grad_norm": 37.28609085083008, + "kl": 0.14453125, + "learning_rate": 7.60038240917782e-07, + "loss": 0.0058, + "reward": 1.4650516510009766, + "reward_std": 0.05887313932180405, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.46505165100097656, + "rewards/pad": 0.0, + "step": 753 + }, + { + "completion_length": 70.875, + "epoch": 0.2402804333970682, + "grad_norm": 28.794374465942383, + "kl": 0.2431640625, + "learning_rate": 7.597195666029318e-07, + "loss": 0.0097, + "reward": 1.8270920515060425, + "reward_std": 0.11150316894054413, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.7020920515060425, + "rewards/pad": 0.125, + "step": 754 + }, + { + "completion_length": 96.046875, + "epoch": 0.24059910771191842, + "grad_norm": 22.261991500854492, + "kl": 0.11767578125, + "learning_rate": 7.594008922880816e-07, + "loss": 0.0047, + "reward": 1.6866573095321655, + "reward_std": 0.045180194079875946, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6866573095321655, + "rewards/pad": 0.0, + "step": 755 + }, + { + "completion_length": 71.75, + "epoch": 0.24091778202676864, + "grad_norm": 33.53934097290039, + "kl": 0.150390625, + "learning_rate": 7.590822179732314e-07, + "loss": 0.006, + "reward": 1.5973482131958008, + "reward_std": 0.1039232611656189, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.472348153591156, + "rewards/pad": 0.125, + "step": 756 + }, + { + "completion_length": 98.234375, + "epoch": 0.24123645634161886, + "grad_norm": 24.93689727783203, + "kl": 0.1025390625, + "learning_rate": 7.587635436583812e-07, + "loss": 0.0041, + "reward": 1.6780802011489868, + "reward_std": 0.10466856509447098, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5530802607536316, + "step": 757 + }, + { + "completion_length": 72.859375, + "epoch": 0.24155513065646908, + "grad_norm": 33.80509567260742, + "kl": 0.10546875, + "learning_rate": 7.584448693435309e-07, + "loss": 0.0042, + "reward": 1.6861860752105713, + "reward_std": 0.08993145823478699, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.43618613481521606, + "step": 758 + }, + { + "completion_length": 46.296875, + "epoch": 0.2418738049713193, + "grad_norm": 38.505409240722656, + "kl": 0.1767578125, + "learning_rate": 7.581261950286807e-07, + "loss": 0.0071, + "reward": 1.5051474571228027, + "reward_std": 0.11469937860965729, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3801473379135132, + "rewards/pad": 0.125, + "step": 759 + }, + { + "completion_length": 150.734375, + "epoch": 0.24219247928616955, + "grad_norm": 51.90461349487305, + "kl": 0.091796875, + "learning_rate": 7.578075207138305e-07, + "loss": 0.0037, + "reward": 1.5936484336853027, + "reward_std": 0.08334853500127792, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.46864843368530273, + "step": 760 + }, + { + "completion_length": 20.125, + "epoch": 0.24251115360101977, + "grad_norm": 49.98828887939453, + "kl": 0.18359375, + "learning_rate": 7.574888463989803e-07, + "loss": 0.0073, + "reward": 1.5463662147521973, + "reward_std": 0.11250153183937073, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4213660955429077, + "rewards/pad": 0.125, + "step": 761 + }, + { + "completion_length": 175.65625, + "epoch": 0.24282982791587, + "grad_norm": 10.89474868774414, + "kl": 0.058349609375, + "learning_rate": 7.5717017208413e-07, + "loss": 0.0023, + "reward": 1.4550466537475586, + "reward_std": 0.05955340713262558, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.45504674315452576, + "step": 762 + }, + { + "completion_length": 97.828125, + "epoch": 0.2431485022307202, + "grad_norm": 26.39861488342285, + "kl": 0.47265625, + "learning_rate": 7.568514977692798e-07, + "loss": 0.0189, + "reward": 1.3110451698303223, + "reward_std": 0.08869006484746933, + "rewards/pad": 0.015625, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.29542022943496704, + "step": 763 + }, + { + "completion_length": 71.5625, + "epoch": 0.24346717654557043, + "grad_norm": 74.72175598144531, + "kl": 0.13671875, + "learning_rate": 7.565328234544296e-07, + "loss": 0.0055, + "reward": 1.6570360660552979, + "reward_std": 0.06887185573577881, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4070361256599426, + "step": 764 + }, + { + "completion_length": 122.078125, + "epoch": 0.24378585086042065, + "grad_norm": 26.987380981445312, + "kl": 0.173828125, + "learning_rate": 7.562141491395793e-07, + "loss": 0.0069, + "reward": 1.4231479167938232, + "reward_std": 0.06835635006427765, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.423147976398468, + "step": 765 + }, + { + "completion_length": 123.5, + "epoch": 0.24410452517527087, + "grad_norm": 18.751310348510742, + "kl": 0.091796875, + "learning_rate": 7.55895474824729e-07, + "loss": 0.0037, + "reward": 1.412963628768921, + "reward_std": 0.04932909086346626, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4129636585712433, + "rewards/pad": 0.0, + "step": 766 + }, + { + "completion_length": 71.390625, + "epoch": 0.2444231994901211, + "grad_norm": 201.6680450439453, + "kl": 0.146484375, + "learning_rate": 7.555768005098788e-07, + "loss": 0.0059, + "reward": 1.555466890335083, + "reward_std": 0.06505988538265228, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.43046680092811584, + "rewards/pad": 0.125, + "step": 767 + }, + { + "completion_length": 97.40625, + "epoch": 0.2447418738049713, + "grad_norm": 33.423675537109375, + "kl": 0.1796875, + "learning_rate": 7.552581261950286e-07, + "loss": 0.0072, + "reward": 1.3956812620162964, + "reward_std": 0.07769222557544708, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.395681232213974, + "rewards/pad": 0.0, + "step": 768 + }, + { + "completion_length": 77.265625, + "epoch": 0.24506054811982153, + "grad_norm": 28.018936157226562, + "kl": 0.1103515625, + "learning_rate": 7.549394518801783e-07, + "loss": 0.0044, + "reward": 1.811413049697876, + "reward_std": 0.07297977805137634, + "rewards/answer_reward": 0.375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.43641290068626404, + "step": 769 + }, + { + "completion_length": 122.46875, + "epoch": 0.24537922243467178, + "grad_norm": 21.377849578857422, + "kl": 0.095703125, + "learning_rate": 7.546207775653281e-07, + "loss": 0.0038, + "reward": 1.536658525466919, + "reward_std": 0.14788584411144257, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.4741585850715637, + "rewards/pad": 0.078125, + "step": 770 + }, + { + "completion_length": 72.6875, + "epoch": 0.245697896749522, + "grad_norm": 68.76750946044922, + "kl": 0.15234375, + "learning_rate": 7.543021032504779e-07, + "loss": 0.0061, + "reward": 1.7531397342681885, + "reward_std": 0.08184407651424408, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5031397938728333, + "rewards/pad": 0.25, + "step": 771 + }, + { + "completion_length": 43.765625, + "epoch": 0.24601657106437222, + "grad_norm": 76.53553771972656, + "kl": 0.16015625, + "learning_rate": 7.539834289356277e-07, + "loss": 0.0064, + "reward": 1.7371867895126343, + "reward_std": 0.11019481718540192, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6121867895126343, + "rewards/pad": 0.125, + "step": 772 + }, + { + "completion_length": 98.421875, + "epoch": 0.24633524537922244, + "grad_norm": 51.312599182128906, + "kl": 0.2119140625, + "learning_rate": 7.536647546207775e-07, + "loss": 0.0085, + "reward": 1.6053152084350586, + "reward_std": 0.05878560245037079, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.48031532764434814, + "rewards/pad": 0.125, + "step": 773 + }, + { + "completion_length": 123.9375, + "epoch": 0.24665391969407266, + "grad_norm": 9.957165718078613, + "kl": 0.16015625, + "learning_rate": 7.533460803059273e-07, + "loss": 0.0064, + "reward": 1.5390030145645142, + "reward_std": 0.054607875645160675, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5390030145645142, + "rewards/pad": 0.0, + "step": 774 + }, + { + "completion_length": 69.703125, + "epoch": 0.24697259400892288, + "grad_norm": 35.7075080871582, + "kl": 0.1162109375, + "learning_rate": 7.530274059910771e-07, + "loss": 0.0046, + "reward": 1.5532881021499634, + "reward_std": 0.06243626028299332, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5532881021499634, + "rewards/pad": 0.0, + "step": 775 + }, + { + "completion_length": 43.390625, + "epoch": 0.2472912683237731, + "grad_norm": 34.16914367675781, + "kl": 0.16796875, + "learning_rate": 7.527087316762269e-07, + "loss": 0.0067, + "reward": 1.642035961151123, + "reward_std": 0.12714436650276184, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5951609015464783, + "rewards/pad": 0.046875, + "step": 776 + }, + { + "completion_length": 96.9375, + "epoch": 0.24760994263862332, + "grad_norm": 64.47351837158203, + "kl": 0.150390625, + "learning_rate": 7.523900573613766e-07, + "loss": 0.006, + "reward": 1.5337800979614258, + "reward_std": 0.11536523699760437, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.533780038356781, + "rewards/pad": 0.0, + "step": 777 + }, + { + "completion_length": 45.4375, + "epoch": 0.24792861695347354, + "grad_norm": 20.62556266784668, + "kl": 0.2265625, + "learning_rate": 7.520713830465264e-07, + "loss": 0.0091, + "reward": 1.589827537536621, + "reward_std": 0.08547364175319672, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5898275375366211, + "rewards/pad": 0.0, + "step": 778 + }, + { + "completion_length": 125.1875, + "epoch": 0.24824729126832376, + "grad_norm": 43.06966018676758, + "kl": 0.25390625, + "learning_rate": 7.517527087316762e-07, + "loss": 0.0101, + "reward": 1.522845983505249, + "reward_std": 0.06987912952899933, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5228460431098938, + "rewards/pad": 0.0, + "step": 779 + }, + { + "completion_length": 126.515625, + "epoch": 0.248565965583174, + "grad_norm": 93.59717559814453, + "kl": 0.10205078125, + "learning_rate": 7.51434034416826e-07, + "loss": 0.0041, + "reward": 1.511330485343933, + "reward_std": 0.07540356367826462, + "rewards/answer_reward": 0.21875, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.2925805449485779, + "step": 780 + }, + { + "completion_length": 96.0625, + "epoch": 0.24888463989802423, + "grad_norm": 24.386804580688477, + "kl": 0.1279296875, + "learning_rate": 7.511153601019757e-07, + "loss": 0.0051, + "reward": 1.5739428997039795, + "reward_std": 0.04993325099349022, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4489428997039795, + "rewards/pad": 0.125, + "step": 781 + }, + { + "completion_length": 97.515625, + "epoch": 0.24920331421287445, + "grad_norm": 21.700822830200195, + "kl": 0.1279296875, + "learning_rate": 7.507966857871255e-07, + "loss": 0.0051, + "reward": 1.456433653831482, + "reward_std": 0.060126226395368576, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.45643362402915955, + "rewards/pad": 0.0, + "step": 782 + }, + { + "completion_length": 72.609375, + "epoch": 0.24952198852772467, + "grad_norm": 12.901528358459473, + "kl": 0.12890625, + "learning_rate": 7.504780114722753e-07, + "loss": 0.0052, + "reward": 1.8618481159210205, + "reward_std": 0.06517429649829865, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.6118481159210205, + "step": 783 + }, + { + "completion_length": 70.9375, + "epoch": 0.2498406628425749, + "grad_norm": 112.09280395507812, + "kl": 0.1435546875, + "learning_rate": 7.501593371574251e-07, + "loss": 0.0057, + "reward": 1.5489466190338135, + "reward_std": 0.061221715062856674, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4239466190338135, + "rewards/pad": 0.125, + "step": 784 + }, + { + "completion_length": 97.859375, + "epoch": 0.2501593371574251, + "grad_norm": 24.970352172851562, + "kl": 0.146484375, + "learning_rate": 7.498406628425748e-07, + "loss": 0.0058, + "reward": 1.50575590133667, + "reward_std": 0.07637608051300049, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5057559609413147, + "rewards/pad": 0.0, + "step": 785 + }, + { + "completion_length": 68.609375, + "epoch": 0.25047801147227533, + "grad_norm": 78.99993133544922, + "kl": 0.11669921875, + "learning_rate": 7.495219885277246e-07, + "loss": 0.0047, + "reward": 1.6616427898406982, + "reward_std": 0.11297961324453354, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.599142849445343, + "rewards/pad": 0.0625, + "step": 786 + }, + { + "completion_length": 96.703125, + "epoch": 0.25079668578712555, + "grad_norm": 28.041793823242188, + "kl": 0.107421875, + "learning_rate": 7.492033142128744e-07, + "loss": 0.0043, + "reward": 1.4415643215179443, + "reward_std": 0.05869434028863907, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.44156432151794434, + "step": 787 + }, + { + "completion_length": 176.875, + "epoch": 0.25111536010197577, + "grad_norm": 38.69453430175781, + "kl": 0.078125, + "learning_rate": 7.488846398980242e-07, + "loss": 0.0031, + "reward": 1.4662871360778809, + "reward_std": 0.04408044368028641, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.466287225484848, + "rewards/pad": 0.0, + "step": 788 + }, + { + "completion_length": 98.40625, + "epoch": 0.251434034416826, + "grad_norm": 77.54596710205078, + "kl": 0.0947265625, + "learning_rate": 7.485659655831739e-07, + "loss": 0.0038, + "reward": 1.611450433731079, + "reward_std": 0.051737017929553986, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4864504635334015, + "rewards/pad": 0.125, + "step": 789 + }, + { + "completion_length": 124.15625, + "epoch": 0.2517527087316762, + "grad_norm": 23.194894790649414, + "kl": 0.07080078125, + "learning_rate": 7.482472912683237e-07, + "loss": 0.0028, + "reward": 1.4636067152023315, + "reward_std": 0.04804643988609314, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.33860671520233154, + "step": 790 + }, + { + "completion_length": 123.953125, + "epoch": 0.25207138304652643, + "grad_norm": 40.468902587890625, + "kl": 0.15234375, + "learning_rate": 7.479286169534736e-07, + "loss": 0.0061, + "reward": 1.4913312196731567, + "reward_std": 0.1270146518945694, + "rewards/pad": 0.015625, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4757062494754791, + "step": 791 + }, + { + "completion_length": 18.625, + "epoch": 0.25239005736137665, + "grad_norm": 27.677337646484375, + "kl": 0.255859375, + "learning_rate": 7.476099426386234e-07, + "loss": 0.0103, + "reward": 1.3575646877288818, + "reward_std": 0.057783663272857666, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.35756468772888184, + "rewards/pad": 0.0, + "step": 792 + }, + { + "completion_length": 122.171875, + "epoch": 0.25270873167622687, + "grad_norm": 113.33208465576172, + "kl": 0.11572265625, + "learning_rate": 7.472912683237731e-07, + "loss": 0.0046, + "reward": 1.457079529762268, + "reward_std": 0.05201934278011322, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.45707952976226807, + "step": 793 + }, + { + "completion_length": 126.421875, + "epoch": 0.2530274059910771, + "grad_norm": 24.9990177154541, + "kl": 0.10498046875, + "learning_rate": 7.469725940089229e-07, + "loss": 0.0042, + "reward": 1.466892957687378, + "reward_std": 0.11991224437952042, + "rewards/answer_reward": 0.09375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.37314292788505554, + "step": 794 + }, + { + "completion_length": 151.875, + "epoch": 0.25334608030592737, + "grad_norm": 20.538066864013672, + "kl": 0.08056640625, + "learning_rate": 7.466539196940727e-07, + "loss": 0.0032, + "reward": 1.6273270845413208, + "reward_std": 0.061867453157901764, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.3773270845413208, + "step": 795 + }, + { + "completion_length": 121.9375, + "epoch": 0.2536647546207776, + "grad_norm": 58.47627639770508, + "kl": 0.11474609375, + "learning_rate": 7.463352453792225e-07, + "loss": 0.0046, + "reward": 1.5742998123168945, + "reward_std": 0.03341483324766159, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5742998719215393, + "rewards/pad": 0.0, + "step": 796 + }, + { + "completion_length": 122.8125, + "epoch": 0.2539834289356278, + "grad_norm": 19.176841735839844, + "kl": 0.130859375, + "learning_rate": 7.460165710643722e-07, + "loss": 0.0052, + "reward": 1.4974907636642456, + "reward_std": 0.04454575479030609, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.49749070405960083, + "rewards/pad": 0.0, + "step": 797 + }, + { + "completion_length": 46.171875, + "epoch": 0.25430210325047803, + "grad_norm": 24.082975387573242, + "kl": 0.1884765625, + "learning_rate": 7.45697896749522e-07, + "loss": 0.0075, + "reward": 1.5651308298110962, + "reward_std": 0.09099891781806946, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4401308000087738, + "step": 798 + }, + { + "completion_length": 72.25, + "epoch": 0.25462077756532825, + "grad_norm": 53.692752838134766, + "kl": 0.1220703125, + "learning_rate": 7.453792224346718e-07, + "loss": 0.0049, + "reward": 1.537758231163025, + "reward_std": 0.09440574795007706, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4283832907676697, + "rewards/pad": 0.109375, + "step": 799 + }, + { + "completion_length": 96.328125, + "epoch": 0.25493945188017847, + "grad_norm": 74.38345336914062, + "kl": 0.1123046875, + "learning_rate": 7.450605481198216e-07, + "loss": 0.0045, + "reward": 1.6785751581192017, + "reward_std": 0.11030935496091843, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5535751581192017, + "rewards/pad": 0.125, + "step": 800 + }, + { + "completion_length": 95.84375, + "epoch": 0.2552581261950287, + "grad_norm": 30.01809310913086, + "kl": 0.123046875, + "learning_rate": 7.447418738049713e-07, + "loss": 0.0049, + "reward": 1.506225347518921, + "reward_std": 0.08522427827119827, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5062253475189209, + "rewards/pad": 0.0, + "step": 801 + }, + { + "completion_length": 68.71875, + "epoch": 0.2555768005098789, + "grad_norm": 36.41709899902344, + "kl": 0.30078125, + "learning_rate": 7.444231994901211e-07, + "loss": 0.012, + "reward": 1.4630727767944336, + "reward_std": 0.0720161646604538, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.463072806596756, + "rewards/pad": 0.0, + "step": 802 + }, + { + "completion_length": 71.28125, + "epoch": 0.25589547482472913, + "grad_norm": 70.04369354248047, + "kl": 0.1279296875, + "learning_rate": 7.441045251752709e-07, + "loss": 0.0051, + "reward": 1.5551037788391113, + "reward_std": 0.09400936961174011, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.43010368943214417, + "step": 803 + }, + { + "completion_length": 122.703125, + "epoch": 0.25621414913957935, + "grad_norm": 89.90359497070312, + "kl": 0.109375, + "learning_rate": 7.437858508604206e-07, + "loss": 0.0044, + "reward": 1.509391188621521, + "reward_std": 0.08077703416347504, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.509391188621521, + "rewards/pad": 0.0, + "step": 804 + }, + { + "completion_length": 72.671875, + "epoch": 0.25653282345442957, + "grad_norm": 76.97059631347656, + "kl": 0.12353515625, + "learning_rate": 7.434671765455703e-07, + "loss": 0.0049, + "reward": 1.4971883296966553, + "reward_std": 0.14855819940567017, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.38781341910362244, + "rewards/pad": 0.125, + "step": 805 + }, + { + "completion_length": 96.84375, + "epoch": 0.2568514977692798, + "grad_norm": 20.679262161254883, + "kl": 0.1884765625, + "learning_rate": 7.431485022307201e-07, + "loss": 0.0075, + "reward": 1.5170658826828003, + "reward_std": 0.07145309448242188, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5170659422874451, + "step": 806 + }, + { + "completion_length": 152.4375, + "epoch": 0.25717017208413, + "grad_norm": 15.460671424865723, + "kl": 0.062255859375, + "learning_rate": 7.428298279158699e-07, + "loss": 0.0025, + "reward": 1.4074444770812988, + "reward_std": 0.09351341426372528, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.423069566488266, + "step": 807 + }, + { + "completion_length": 227.65625, + "epoch": 0.25748884639898023, + "grad_norm": 1.5667763948440552, + "kl": 0.0279541015625, + "learning_rate": 7.425111536010196e-07, + "loss": 0.0011, + "reward": 1.2311536073684692, + "reward_std": 0.08234579861164093, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.24677860736846924, + "step": 808 + }, + { + "completion_length": 45.359375, + "epoch": 0.25780752071383045, + "grad_norm": 122.13186645507812, + "kl": 0.1298828125, + "learning_rate": 7.421924792861694e-07, + "loss": 0.0052, + "reward": 1.8012471199035645, + "reward_std": 0.09172539412975311, + "rewards/answer_reward": 0.375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.42624717950820923, + "step": 809 + }, + { + "completion_length": 96.296875, + "epoch": 0.25812619502868067, + "grad_norm": 26.060197830200195, + "kl": 0.11181640625, + "learning_rate": 7.418738049713192e-07, + "loss": 0.0045, + "reward": 1.5975518226623535, + "reward_std": 0.08361221849918365, + "rewards/pad": 0.09375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5038020014762878, + "step": 810 + }, + { + "completion_length": 98.15625, + "epoch": 0.2584448693435309, + "grad_norm": 31.350692749023438, + "kl": 0.134765625, + "learning_rate": 7.415551306564691e-07, + "loss": 0.0054, + "reward": 1.492019772529602, + "reward_std": 0.0747697651386261, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.49201980233192444, + "rewards/pad": 0.0, + "step": 811 + }, + { + "completion_length": 71.140625, + "epoch": 0.2587635436583811, + "grad_norm": 73.38872528076172, + "kl": 0.125, + "learning_rate": 7.412364563416188e-07, + "loss": 0.005, + "reward": 1.751183271408081, + "reward_std": 0.06803754717111588, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.626183271408081, + "step": 812 + }, + { + "completion_length": 46.9375, + "epoch": 0.25908221797323133, + "grad_norm": 83.22901153564453, + "kl": 0.16796875, + "learning_rate": 7.409177820267686e-07, + "loss": 0.0067, + "reward": 1.6288355588912964, + "reward_std": 0.1803748458623886, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5194604992866516, + "rewards/pad": 0.109375, + "step": 813 + }, + { + "completion_length": 72.671875, + "epoch": 0.2594008922880816, + "grad_norm": 153.54087829589844, + "kl": 0.13671875, + "learning_rate": 7.405991077119184e-07, + "loss": 0.0055, + "reward": 1.5559117794036865, + "reward_std": 0.13691410422325134, + "rewards/pad": 0.171875, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3840367794036865, + "step": 814 + }, + { + "completion_length": 98.59375, + "epoch": 0.25971956660293183, + "grad_norm": 33.62237548828125, + "kl": 0.1328125, + "learning_rate": 7.402804333970682e-07, + "loss": 0.0053, + "reward": 1.7688663005828857, + "reward_std": 0.04326247423887253, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.643866240978241, + "rewards/pad": 0.125, + "step": 815 + }, + { + "completion_length": 172.515625, + "epoch": 0.26003824091778205, + "grad_norm": 67.81851959228516, + "kl": 0.0830078125, + "learning_rate": 7.399617590822179e-07, + "loss": 0.0033, + "reward": 1.5516440868377686, + "reward_std": 0.05654784291982651, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.426643967628479, + "step": 816 + }, + { + "completion_length": 149.671875, + "epoch": 0.26035691523263227, + "grad_norm": 25.504825592041016, + "kl": 0.07763671875, + "learning_rate": 7.396430847673677e-07, + "loss": 0.0031, + "reward": 1.3726608753204346, + "reward_std": 0.02660282514989376, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.37266087532043457, + "step": 817 + }, + { + "completion_length": 71.375, + "epoch": 0.2606755895474825, + "grad_norm": 20.733476638793945, + "kl": 0.228515625, + "learning_rate": 7.393244104525175e-07, + "loss": 0.0091, + "reward": 1.8035805225372314, + "reward_std": 0.10151496529579163, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5535804629325867, + "rewards/pad": 0.25, + "step": 818 + }, + { + "completion_length": 97.46875, + "epoch": 0.2609942638623327, + "grad_norm": 45.76640319824219, + "kl": 0.16015625, + "learning_rate": 7.390057361376673e-07, + "loss": 0.0064, + "reward": 1.468562364578247, + "reward_std": 0.04567372053861618, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4685623049736023, + "rewards/pad": 0.0, + "step": 819 + }, + { + "completion_length": 101.09375, + "epoch": 0.26131293817718293, + "grad_norm": 44.104248046875, + "kl": 0.142578125, + "learning_rate": 7.38687061822817e-07, + "loss": 0.0057, + "reward": 1.7732608318328857, + "reward_std": 0.03834523260593414, + "rewards/pad": 0.375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3982608914375305, + "step": 820 + }, + { + "completion_length": 100.40625, + "epoch": 0.26163161249203315, + "grad_norm": 84.38065338134766, + "kl": 0.09033203125, + "learning_rate": 7.383683875079668e-07, + "loss": 0.0036, + "reward": 1.8088653087615967, + "reward_std": 0.12013958394527435, + "rewards/pad": 0.3125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4963652491569519, + "step": 821 + }, + { + "completion_length": 69.9375, + "epoch": 0.26195028680688337, + "grad_norm": 61.65689468383789, + "kl": 0.25390625, + "learning_rate": 7.380497131931166e-07, + "loss": 0.0102, + "reward": 1.5655962228775024, + "reward_std": 0.07943125814199448, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.44059622287750244, + "step": 822 + }, + { + "completion_length": 96.5625, + "epoch": 0.2622689611217336, + "grad_norm": 35.45212936401367, + "kl": 0.3828125, + "learning_rate": 7.377310388782664e-07, + "loss": 0.0153, + "reward": 1.6489589214324951, + "reward_std": 0.035216521471738815, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6489588618278503, + "rewards/pad": 0.0, + "step": 823 + }, + { + "completion_length": 151.34375, + "epoch": 0.2625876354365838, + "grad_norm": 18.12538719177246, + "kl": 0.0634765625, + "learning_rate": 7.374123645634161e-07, + "loss": 0.0025, + "reward": 1.692170262336731, + "reward_std": 0.0424518883228302, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5671701431274414, + "step": 824 + }, + { + "completion_length": 74.03125, + "epoch": 0.26290630975143403, + "grad_norm": 138.32925415039062, + "kl": 0.1396484375, + "learning_rate": 7.370936902485659e-07, + "loss": 0.0056, + "reward": 1.632441759109497, + "reward_std": 0.0912943109869957, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5074419379234314, + "rewards/pad": 0.125, + "step": 825 + }, + { + "completion_length": 126.109375, + "epoch": 0.26322498406628425, + "grad_norm": 28.759326934814453, + "kl": 0.10205078125, + "learning_rate": 7.367750159337157e-07, + "loss": 0.0041, + "reward": 1.5843374729156494, + "reward_std": 0.06639618426561356, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.45933741331100464, + "step": 826 + }, + { + "completion_length": 96.734375, + "epoch": 0.26354365838113447, + "grad_norm": 58.95811462402344, + "kl": 0.107421875, + "learning_rate": 7.364563416188655e-07, + "loss": 0.0043, + "reward": 1.32325279712677, + "reward_std": 0.03803478926420212, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3232528269290924, + "rewards/pad": 0.0, + "step": 827 + }, + { + "completion_length": 98.109375, + "epoch": 0.2638623326959847, + "grad_norm": 28.35196304321289, + "kl": 0.1337890625, + "learning_rate": 7.361376673040152e-07, + "loss": 0.0054, + "reward": 1.5631272792816162, + "reward_std": 0.04592440277338028, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5631272792816162, + "rewards/pad": 0.0, + "step": 828 + }, + { + "completion_length": 44.65625, + "epoch": 0.2641810070108349, + "grad_norm": 42.16221237182617, + "kl": 0.1748046875, + "learning_rate": 7.35818992989165e-07, + "loss": 0.007, + "reward": 1.7024767398834229, + "reward_std": 0.05674516409635544, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5774767994880676, + "rewards/pad": 0.125, + "step": 829 + }, + { + "completion_length": 176.4375, + "epoch": 0.26449968132568513, + "grad_norm": 92.30027770996094, + "kl": 0.08935546875, + "learning_rate": 7.355003186743149e-07, + "loss": 0.0036, + "reward": 1.4316385984420776, + "reward_std": 0.10597049444913864, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.32226359844207764, + "rewards/pad": 0.125, + "step": 830 + }, + { + "completion_length": 146.125, + "epoch": 0.26481835564053535, + "grad_norm": 36.19331359863281, + "kl": 0.126953125, + "learning_rate": 7.351816443594647e-07, + "loss": 0.0051, + "reward": 1.4525105953216553, + "reward_std": 0.08761756867170334, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4525105357170105, + "rewards/pad": 0.0, + "step": 831 + }, + { + "completion_length": 99.609375, + "epoch": 0.2651370299553856, + "grad_norm": 43.91147232055664, + "kl": 0.14453125, + "learning_rate": 7.348629700446144e-07, + "loss": 0.0058, + "reward": 1.6105023622512817, + "reward_std": 0.0791240856051445, + "rewards/pad": 0.109375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5011274218559265, + "step": 832 + }, + { + "completion_length": 150.46875, + "epoch": 0.2654557042702358, + "grad_norm": 13.669109344482422, + "kl": 0.10302734375, + "learning_rate": 7.345442957297642e-07, + "loss": 0.0041, + "reward": 1.4903199672698975, + "reward_std": 0.0454377755522728, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4903199076652527, + "rewards/pad": 0.0, + "step": 833 + }, + { + "completion_length": 45.953125, + "epoch": 0.26577437858508607, + "grad_norm": 64.69865417480469, + "kl": 0.203125, + "learning_rate": 7.34225621414914e-07, + "loss": 0.0081, + "reward": 1.657322883605957, + "reward_std": 0.054739683866500854, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.532322883605957, + "step": 834 + }, + { + "completion_length": 45.140625, + "epoch": 0.2660930528999363, + "grad_norm": 32.49448776245117, + "kl": 0.1376953125, + "learning_rate": 7.339069471000637e-07, + "loss": 0.0055, + "reward": 1.6843942403793335, + "reward_std": 0.05278486758470535, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.559394121170044, + "rewards/pad": 0.125, + "step": 835 + }, + { + "completion_length": 150.9375, + "epoch": 0.2664117272147865, + "grad_norm": 10.046256065368652, + "kl": 0.06298828125, + "learning_rate": 7.335882727852135e-07, + "loss": 0.0025, + "reward": 1.5606993436813354, + "reward_std": 0.057660944759845734, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.43569934368133545, + "step": 836 + }, + { + "completion_length": 95.84375, + "epoch": 0.26673040152963673, + "grad_norm": 21.493789672851562, + "kl": 0.16796875, + "learning_rate": 7.332695984703633e-07, + "loss": 0.0067, + "reward": 1.5020877122879028, + "reward_std": 0.053312573581933975, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5020877718925476, + "step": 837 + }, + { + "completion_length": 127.125, + "epoch": 0.26704907584448695, + "grad_norm": 40.390464782714844, + "kl": 0.08154296875, + "learning_rate": 7.329509241555131e-07, + "loss": 0.0033, + "reward": 1.6771814823150635, + "reward_std": 0.05436215549707413, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.42718154191970825, + "rewards/pad": 0.25, + "step": 838 + }, + { + "completion_length": 97.796875, + "epoch": 0.26736775015933717, + "grad_norm": 33.54790115356445, + "kl": 0.12890625, + "learning_rate": 7.326322498406628e-07, + "loss": 0.0051, + "reward": 1.6265747547149658, + "reward_std": 0.031377844512462616, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6265748143196106, + "rewards/pad": 0.0, + "step": 839 + }, + { + "completion_length": 125.875, + "epoch": 0.2676864244741874, + "grad_norm": 120.50456237792969, + "kl": 0.0810546875, + "learning_rate": 7.323135755258126e-07, + "loss": 0.0032, + "reward": 1.562516212463379, + "reward_std": 0.11323312669992447, + "rewards/answer_reward": 0.140625, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.42189115285873413, + "step": 840 + }, + { + "completion_length": 97.3125, + "epoch": 0.2680050987890376, + "grad_norm": 15.057744026184082, + "kl": 0.1298828125, + "learning_rate": 7.319949012109624e-07, + "loss": 0.0052, + "reward": 1.3758224248886108, + "reward_std": 0.02350495010614395, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.37582242488861084, + "step": 841 + }, + { + "completion_length": 72.015625, + "epoch": 0.26832377310388783, + "grad_norm": 23.781997680664062, + "kl": 0.2236328125, + "learning_rate": 7.316762268961122e-07, + "loss": 0.0089, + "reward": 1.5852105617523193, + "reward_std": 0.09232565760612488, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4758354425430298, + "rewards/pad": 0.109375, + "step": 842 + }, + { + "completion_length": 73.34375, + "epoch": 0.26864244741873805, + "grad_norm": 40.67388916015625, + "kl": 0.1728515625, + "learning_rate": 7.313575525812619e-07, + "loss": 0.0069, + "reward": 1.6226885318756104, + "reward_std": 0.08350202441215515, + "rewards/answer_reward": 0.234375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.3883134126663208, + "step": 843 + }, + { + "completion_length": 72.21875, + "epoch": 0.26896112173358827, + "grad_norm": 62.01053237915039, + "kl": 0.12255859375, + "learning_rate": 7.310388782664116e-07, + "loss": 0.0049, + "reward": 1.8127069473266602, + "reward_std": 0.07875151187181473, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5627070069313049, + "step": 844 + }, + { + "completion_length": 96.5625, + "epoch": 0.2692797960484385, + "grad_norm": 26.037363052368164, + "kl": 0.2177734375, + "learning_rate": 7.307202039515614e-07, + "loss": 0.0087, + "reward": 1.4820842742919922, + "reward_std": 0.0368165522813797, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.482084184885025, + "rewards/pad": 0.0, + "step": 845 + }, + { + "completion_length": 148.109375, + "epoch": 0.2695984703632887, + "grad_norm": 32.60529708862305, + "kl": 0.0791015625, + "learning_rate": 7.304015296367112e-07, + "loss": 0.0032, + "reward": 1.5095293521881104, + "reward_std": 0.07287216186523438, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5095293521881104, + "rewards/pad": 0.0, + "step": 846 + }, + { + "completion_length": 96.15625, + "epoch": 0.26991714467813893, + "grad_norm": 40.325313568115234, + "kl": 0.171875, + "learning_rate": 7.300828553218609e-07, + "loss": 0.0069, + "reward": 1.5568125247955322, + "reward_std": 0.07620823383331299, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5568124651908875, + "step": 847 + }, + { + "completion_length": 123.203125, + "epoch": 0.27023581899298915, + "grad_norm": 72.05610656738281, + "kl": 0.1494140625, + "learning_rate": 7.297641810070108e-07, + "loss": 0.006, + "reward": 1.379019021987915, + "reward_std": 0.04552176594734192, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3790189027786255, + "step": 848 + }, + { + "completion_length": 71.984375, + "epoch": 0.27055449330783937, + "grad_norm": 31.67304039001465, + "kl": 0.1259765625, + "learning_rate": 7.294455066921606e-07, + "loss": 0.005, + "reward": 1.5310370922088623, + "reward_std": 0.044179461896419525, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4060371518135071, + "step": 849 + }, + { + "completion_length": 70.65625, + "epoch": 0.2708731676226896, + "grad_norm": 36.889957427978516, + "kl": 0.1513671875, + "learning_rate": 7.291268323773104e-07, + "loss": 0.0061, + "reward": 1.602161169052124, + "reward_std": 0.11161001771688461, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6021612286567688, + "rewards/pad": 0.0, + "step": 850 + }, + { + "completion_length": 96.84375, + "epoch": 0.2711918419375398, + "grad_norm": 35.18388366699219, + "kl": 0.1357421875, + "learning_rate": 7.288081580624601e-07, + "loss": 0.0054, + "reward": 1.5036709308624268, + "reward_std": 0.06766320019960403, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5036708116531372, + "step": 851 + }, + { + "completion_length": 93.984375, + "epoch": 0.27151051625239003, + "grad_norm": 127.4382553100586, + "kl": 1.5390625, + "learning_rate": 7.284894837476099e-07, + "loss": 0.0615, + "reward": 1.473995566368103, + "reward_std": 0.09904636442661285, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.36462053656578064, + "rewards/pad": 0.125, + "step": 852 + }, + { + "completion_length": 69.3125, + "epoch": 0.27182919056724025, + "grad_norm": 42.15987777709961, + "kl": 0.1513671875, + "learning_rate": 7.281708094327597e-07, + "loss": 0.0061, + "reward": 1.719491958618164, + "reward_std": 0.05202781781554222, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.7194918990135193, + "rewards/pad": 0.0, + "step": 853 + }, + { + "completion_length": 44.765625, + "epoch": 0.27214786488209053, + "grad_norm": 56.77442169189453, + "kl": 0.2177734375, + "learning_rate": 7.278521351179095e-07, + "loss": 0.0087, + "reward": 1.592527151107788, + "reward_std": 0.03890862315893173, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5925271511077881, + "rewards/pad": 0.0, + "step": 854 + }, + { + "completion_length": 71.765625, + "epoch": 0.27246653919694075, + "grad_norm": 78.89741516113281, + "kl": 0.2138671875, + "learning_rate": 7.275334608030592e-07, + "loss": 0.0086, + "reward": 1.6108283996582031, + "reward_std": 0.14424830675125122, + "rewards/pad": 0.078125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5327032804489136, + "step": 855 + }, + { + "completion_length": 71.15625, + "epoch": 0.27278521351179097, + "grad_norm": 23.657489776611328, + "kl": 0.298828125, + "learning_rate": 7.27214786488209e-07, + "loss": 0.012, + "reward": 1.430689811706543, + "reward_std": 0.0537835955619812, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.43068990111351013, + "rewards/pad": 0.0, + "step": 856 + }, + { + "completion_length": 44.703125, + "epoch": 0.2731038878266412, + "grad_norm": 31.861421585083008, + "kl": 0.271484375, + "learning_rate": 7.268961121733588e-07, + "loss": 0.0108, + "reward": 1.5638333559036255, + "reward_std": 0.06311056017875671, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5638333559036255, + "step": 857 + }, + { + "completion_length": 124.09375, + "epoch": 0.2734225621414914, + "grad_norm": 16.415027618408203, + "kl": 0.10693359375, + "learning_rate": 7.265774378585086e-07, + "loss": 0.0043, + "reward": 1.59592604637146, + "reward_std": 0.05373538285493851, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.59592604637146, + "rewards/pad": 0.0, + "step": 858 + }, + { + "completion_length": 45.921875, + "epoch": 0.27374123645634163, + "grad_norm": 86.9303207397461, + "kl": 0.2490234375, + "learning_rate": 7.262587635436583e-07, + "loss": 0.0099, + "reward": 1.4179108142852783, + "reward_std": 0.07070551812648773, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.2929109036922455, + "rewards/pad": 0.125, + "step": 859 + }, + { + "completion_length": 121.875, + "epoch": 0.27405991077119185, + "grad_norm": 39.70486068725586, + "kl": 0.140625, + "learning_rate": 7.259400892288081e-07, + "loss": 0.0056, + "reward": 1.5685955286026, + "reward_std": 0.04470613971352577, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5685955286026001, + "step": 860 + }, + { + "completion_length": 45.171875, + "epoch": 0.27437858508604207, + "grad_norm": 56.10869598388672, + "kl": 0.150390625, + "learning_rate": 7.256214149139579e-07, + "loss": 0.006, + "reward": 1.7189316749572754, + "reward_std": 0.0561370775103569, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5939316749572754, + "rewards/pad": 0.125, + "step": 861 + }, + { + "completion_length": 96.03125, + "epoch": 0.2746972594008923, + "grad_norm": 57.85542297363281, + "kl": 0.1044921875, + "learning_rate": 7.253027405991076e-07, + "loss": 0.0042, + "reward": 1.52811861038208, + "reward_std": 0.06384322047233582, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5281186103820801, + "rewards/pad": 0.0, + "step": 862 + }, + { + "completion_length": 73.0625, + "epoch": 0.2750159337157425, + "grad_norm": 117.91090393066406, + "kl": 0.1845703125, + "learning_rate": 7.249840662842574e-07, + "loss": 0.0074, + "reward": 1.769296407699585, + "reward_std": 0.07024556398391724, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.519296407699585, + "rewards/pad": 0.25, + "step": 863 + }, + { + "completion_length": 72.734375, + "epoch": 0.27533460803059273, + "grad_norm": 28.344497680664062, + "kl": 0.166015625, + "learning_rate": 7.246653919694072e-07, + "loss": 0.0066, + "reward": 1.793961763381958, + "reward_std": 0.07941886782646179, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.543961763381958, + "rewards/pad": 0.25, + "step": 864 + }, + { + "completion_length": 124.421875, + "epoch": 0.27565328234544295, + "grad_norm": 25.100507736206055, + "kl": 0.0849609375, + "learning_rate": 7.24346717654557e-07, + "loss": 0.0034, + "reward": 1.7098156213760376, + "reward_std": 0.051566511392593384, + "rewards/pad": 0.375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3348156809806824, + "step": 865 + }, + { + "completion_length": 72.515625, + "epoch": 0.27597195666029317, + "grad_norm": 54.915462493896484, + "kl": 0.10302734375, + "learning_rate": 7.240280433397067e-07, + "loss": 0.0041, + "reward": 1.5807334184646606, + "reward_std": 0.09976384043693542, + "rewards/answer_reward": 0.171875, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.40885844826698303, + "step": 866 + }, + { + "completion_length": 121.40625, + "epoch": 0.2762906309751434, + "grad_norm": 19.998388290405273, + "kl": 0.169921875, + "learning_rate": 7.237093690248566e-07, + "loss": 0.0068, + "reward": 1.4986286163330078, + "reward_std": 0.0763099268078804, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.49862855672836304, + "step": 867 + }, + { + "completion_length": 97.390625, + "epoch": 0.2766093052899936, + "grad_norm": 19.720415115356445, + "kl": 0.11572265625, + "learning_rate": 7.233906947100064e-07, + "loss": 0.0046, + "reward": 1.6597509384155273, + "reward_std": 0.04936613887548447, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4097510576248169, + "step": 868 + }, + { + "completion_length": 97.96875, + "epoch": 0.27692797960484383, + "grad_norm": 26.631547927856445, + "kl": 0.134765625, + "learning_rate": 7.230720203951562e-07, + "loss": 0.0054, + "reward": 1.6074879169464111, + "reward_std": 0.05781853571534157, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3574879467487335, + "rewards/pad": 0.25, + "step": 869 + }, + { + "completion_length": 18.40625, + "epoch": 0.27724665391969405, + "grad_norm": 75.75119018554688, + "kl": 0.201171875, + "learning_rate": 7.227533460803059e-07, + "loss": 0.0081, + "reward": 1.5645930767059326, + "reward_std": 0.1367059350013733, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5645930767059326, + "rewards/pad": 0.0, + "step": 870 + }, + { + "completion_length": 45.1875, + "epoch": 0.2775653282345443, + "grad_norm": 40.70870590209961, + "kl": 0.267578125, + "learning_rate": 7.224346717654557e-07, + "loss": 0.0107, + "reward": 1.689696192741394, + "reward_std": 0.05542924627661705, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5646960735321045, + "rewards/pad": 0.125, + "step": 871 + }, + { + "completion_length": 94.90625, + "epoch": 0.2778840025493945, + "grad_norm": 26.831253051757812, + "kl": 0.259765625, + "learning_rate": 7.221159974506055e-07, + "loss": 0.0104, + "reward": 1.406510829925537, + "reward_std": 0.0437244214117527, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4065108895301819, + "rewards/pad": 0.0, + "step": 872 + }, + { + "completion_length": 92.234375, + "epoch": 0.27820267686424477, + "grad_norm": 33.70995330810547, + "kl": 0.171875, + "learning_rate": 7.217973231357553e-07, + "loss": 0.0069, + "reward": 1.5849461555480957, + "reward_std": 0.07326267659664154, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4599460959434509, + "rewards/pad": 0.125, + "step": 873 + }, + { + "completion_length": 98.453125, + "epoch": 0.278521351179095, + "grad_norm": 24.882591247558594, + "kl": 0.1767578125, + "learning_rate": 7.21478648820905e-07, + "loss": 0.0071, + "reward": 1.6644113063812256, + "reward_std": 0.11803185939788818, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.5550362467765808, + "step": 874 + }, + { + "completion_length": 69.03125, + "epoch": 0.2788400254939452, + "grad_norm": 20.49930191040039, + "kl": 0.2734375, + "learning_rate": 7.211599745060548e-07, + "loss": 0.0109, + "reward": 1.5971676111221313, + "reward_std": 0.06809830665588379, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5971676111221313, + "rewards/pad": 0.0, + "step": 875 + }, + { + "completion_length": 114.796875, + "epoch": 0.27915869980879543, + "grad_norm": 31.659582138061523, + "kl": 0.1357421875, + "learning_rate": 7.208413001912046e-07, + "loss": 0.0054, + "reward": 1.5567210912704468, + "reward_std": 0.04574970528483391, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5567210912704468, + "rewards/pad": 0.0, + "step": 876 + }, + { + "completion_length": 121.578125, + "epoch": 0.27947737412364565, + "grad_norm": 13.323171615600586, + "kl": 0.20703125, + "learning_rate": 7.205226258763544e-07, + "loss": 0.0083, + "reward": 1.4455690383911133, + "reward_std": 0.02520911768078804, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4455690085887909, + "rewards/pad": 0.0, + "step": 877 + }, + { + "completion_length": 121.84375, + "epoch": 0.27979604843849587, + "grad_norm": 81.34120178222656, + "kl": 0.1240234375, + "learning_rate": 7.202039515615041e-07, + "loss": 0.005, + "reward": 1.5624570846557617, + "reward_std": 0.19658038020133972, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.4218320846557617, + "rewards/pad": 0.15625, + "step": 878 + }, + { + "completion_length": 120.375, + "epoch": 0.2801147227533461, + "grad_norm": 20.47405242919922, + "kl": 0.11865234375, + "learning_rate": 7.198852772466539e-07, + "loss": 0.0047, + "reward": 1.713339924812317, + "reward_std": 0.04620824754238129, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5883399248123169, + "step": 879 + }, + { + "completion_length": 123.046875, + "epoch": 0.2804333970681963, + "grad_norm": 10.733762741088867, + "kl": 0.15625, + "learning_rate": 7.195666029318037e-07, + "loss": 0.0063, + "reward": 1.3499104976654053, + "reward_std": 0.020909493789076805, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.2249104231595993, + "step": 880 + }, + { + "completion_length": 120.375, + "epoch": 0.28075207138304653, + "grad_norm": 24.233781814575195, + "kl": 0.146484375, + "learning_rate": 7.192479286169535e-07, + "loss": 0.0058, + "reward": 1.4725770950317383, + "reward_std": 0.09297582507133484, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.47257712483406067, + "step": 881 + }, + { + "completion_length": 95.921875, + "epoch": 0.28107074569789675, + "grad_norm": 11.09907054901123, + "kl": 0.1025390625, + "learning_rate": 7.189292543021032e-07, + "loss": 0.0041, + "reward": 1.7779861688613892, + "reward_std": 0.06479617953300476, + "rewards/pad": 0.375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.40298622846603394, + "step": 882 + }, + { + "completion_length": 70.9375, + "epoch": 0.28138942001274697, + "grad_norm": 82.33478546142578, + "kl": 0.1689453125, + "learning_rate": 7.186105799872529e-07, + "loss": 0.0067, + "reward": 1.7681920528411865, + "reward_std": 0.054539378732442856, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5181920528411865, + "rewards/pad": 0.25, + "step": 883 + }, + { + "completion_length": 118.875, + "epoch": 0.2817080943275972, + "grad_norm": 22.16440200805664, + "kl": 0.1201171875, + "learning_rate": 7.182919056724027e-07, + "loss": 0.0048, + "reward": 1.3002312183380127, + "reward_std": 0.03425529599189758, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3002312481403351, + "step": 884 + }, + { + "completion_length": 68.15625, + "epoch": 0.2820267686424474, + "grad_norm": 30.926509857177734, + "kl": 0.34765625, + "learning_rate": 7.179732313575525e-07, + "loss": 0.0139, + "reward": 1.5398776531219482, + "reward_std": 0.03936472162604332, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5398776531219482, + "rewards/pad": 0.0, + "step": 885 + }, + { + "completion_length": 94.421875, + "epoch": 0.28234544295729763, + "grad_norm": 46.35068130493164, + "kl": 0.16796875, + "learning_rate": 7.176545570427023e-07, + "loss": 0.0067, + "reward": 1.3887062072753906, + "reward_std": 0.1097392737865448, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3887063264846802, + "rewards/pad": 0.0, + "step": 886 + }, + { + "completion_length": 69.5, + "epoch": 0.28266411727214785, + "grad_norm": 26.24706268310547, + "kl": 0.1689453125, + "learning_rate": 7.173358827278521e-07, + "loss": 0.0067, + "reward": 1.6714463233947754, + "reward_std": 0.10547646135091782, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.5620713233947754, + "step": 887 + }, + { + "completion_length": 147.4375, + "epoch": 0.2829827915869981, + "grad_norm": 10.859973907470703, + "kl": 0.11279296875, + "learning_rate": 7.170172084130019e-07, + "loss": 0.0045, + "reward": 1.453810691833496, + "reward_std": 0.03332022577524185, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.45381075143814087, + "rewards/pad": 0.0, + "step": 888 + }, + { + "completion_length": 46.78125, + "epoch": 0.2833014659018483, + "grad_norm": 49.23662185668945, + "kl": 0.2734375, + "learning_rate": 7.166985340981517e-07, + "loss": 0.0109, + "reward": 1.8417705297470093, + "reward_std": 0.1342671513557434, + "rewards/answer_reward": 0.34375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4980205297470093, + "step": 889 + }, + { + "completion_length": 70.390625, + "epoch": 0.2836201402166985, + "grad_norm": 36.180294036865234, + "kl": 0.263671875, + "learning_rate": 7.163798597833014e-07, + "loss": 0.0106, + "reward": 1.3964661359786987, + "reward_std": 0.08337822556495667, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.39646613597869873, + "rewards/pad": 0.0, + "step": 890 + }, + { + "completion_length": 94.796875, + "epoch": 0.28393881453154873, + "grad_norm": 146.4197540283203, + "kl": 0.11474609375, + "learning_rate": 7.160611854684512e-07, + "loss": 0.0046, + "reward": 1.4267725944519043, + "reward_std": 0.11456167697906494, + "rewards/answer_reward": 0.0, + "rewards/format_reward_gqa": 0.984375, + "rewards/iou_glue_reward": 0.4423975348472595, + "step": 891 + }, + { + "completion_length": 97.171875, + "epoch": 0.28425748884639895, + "grad_norm": 39.37761688232422, + "kl": 0.1611328125, + "learning_rate": 7.15742511153601e-07, + "loss": 0.0065, + "reward": 1.6899974346160889, + "reward_std": 0.05695369839668274, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5649974346160889, + "step": 892 + }, + { + "completion_length": 21.546875, + "epoch": 0.28457616316124923, + "grad_norm": 43.16127014160156, + "kl": 0.30078125, + "learning_rate": 7.154238368387507e-07, + "loss": 0.012, + "reward": 1.8982596397399902, + "reward_std": 0.11875072121620178, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5388847589492798, + "rewards/pad": 0.359375, + "step": 893 + }, + { + "completion_length": 145.640625, + "epoch": 0.28489483747609945, + "grad_norm": 22.15288543701172, + "kl": 0.091796875, + "learning_rate": 7.151051625239005e-07, + "loss": 0.0037, + "reward": 1.303932785987854, + "reward_std": 0.03641955554485321, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.303932785987854, + "step": 894 + }, + { + "completion_length": 72.515625, + "epoch": 0.28521351179094967, + "grad_norm": 97.79877471923828, + "kl": 0.126953125, + "learning_rate": 7.147864882090503e-07, + "loss": 0.0051, + "reward": 1.45205557346344, + "reward_std": 0.054157327860593796, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3270556330680847, + "rewards/pad": 0.125, + "step": 895 + }, + { + "completion_length": 124.28125, + "epoch": 0.2855321861057999, + "grad_norm": 70.17988586425781, + "kl": 0.0830078125, + "learning_rate": 7.144678138942001e-07, + "loss": 0.0033, + "reward": 1.5320907831192017, + "reward_std": 0.07828199118375778, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.40709084272384644, + "step": 896 + }, + { + "completion_length": 123.953125, + "epoch": 0.2858508604206501, + "grad_norm": 22.958566665649414, + "kl": 0.93359375, + "learning_rate": 7.141491395793498e-07, + "loss": 0.0374, + "reward": 1.4208877086639404, + "reward_std": 0.07943449914455414, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.42088767886161804, + "rewards/pad": 0.0, + "step": 897 + }, + { + "completion_length": 72.609375, + "epoch": 0.28616953473550033, + "grad_norm": 18.273794174194336, + "kl": 0.10595703125, + "learning_rate": 7.138304652644996e-07, + "loss": 0.0042, + "reward": 1.5178472995758057, + "reward_std": 0.12105812132358551, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.40847232937812805, + "rewards/pad": 0.125, + "step": 898 + }, + { + "completion_length": 122.671875, + "epoch": 0.28648820905035055, + "grad_norm": 34.814266204833984, + "kl": 0.1103515625, + "learning_rate": 7.135117909496494e-07, + "loss": 0.0044, + "reward": 1.6072380542755127, + "reward_std": 0.0331178680062294, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.48223793506622314, + "step": 899 + }, + { + "completion_length": 69.875, + "epoch": 0.28680688336520077, + "grad_norm": 83.57713317871094, + "kl": 0.1455078125, + "learning_rate": 7.131931166347992e-07, + "loss": 0.0058, + "reward": 1.6023356914520264, + "reward_std": 0.12720996141433716, + "rewards/answer_reward": 0.140625, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4617108106613159, + "step": 900 + }, + { + "completion_length": 95.203125, + "epoch": 0.287125557680051, + "grad_norm": 64.39884185791016, + "kl": 0.08154296875, + "learning_rate": 7.128744423199489e-07, + "loss": 0.0033, + "reward": 1.5819507837295532, + "reward_std": 0.10809376835823059, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4569507837295532, + "step": 901 + }, + { + "completion_length": 96.875, + "epoch": 0.2874442319949012, + "grad_norm": 21.195661544799805, + "kl": 0.126953125, + "learning_rate": 7.125557680050987e-07, + "loss": 0.0051, + "reward": 1.4954230785369873, + "reward_std": 0.10859449207782745, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.5110480785369873, + "rewards/pad": 0.0, + "step": 902 + }, + { + "completion_length": 173.125, + "epoch": 0.28776290630975143, + "grad_norm": 5.571096420288086, + "kl": 0.06787109375, + "learning_rate": 7.122370936902485e-07, + "loss": 0.0027, + "reward": 1.541063904762268, + "reward_std": 0.12838870286941528, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.43168896436691284, + "step": 903 + }, + { + "completion_length": 45.6875, + "epoch": 0.28808158062460165, + "grad_norm": 67.46749114990234, + "kl": 0.11083984375, + "learning_rate": 7.119184193753984e-07, + "loss": 0.0044, + "reward": 1.7282071113586426, + "reward_std": 0.08071044832468033, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6032070517539978, + "rewards/pad": 0.125, + "step": 904 + }, + { + "completion_length": 147.40625, + "epoch": 0.28840025493945187, + "grad_norm": 16.74541473388672, + "kl": 0.0634765625, + "learning_rate": 7.115997450605481e-07, + "loss": 0.0025, + "reward": 1.4114105701446533, + "reward_std": 0.09488704055547714, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.4270356297492981, + "rewards/pad": 0.0, + "step": 905 + }, + { + "completion_length": 19.859375, + "epoch": 0.2887189292543021, + "grad_norm": 121.58633422851562, + "kl": 2.765625, + "learning_rate": 7.112810707456979e-07, + "loss": 0.1109, + "reward": 1.7666852474212646, + "reward_std": 0.18628299236297607, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.5323102474212646, + "rewards/pad": 0.25, + "step": 906 + }, + { + "completion_length": 120.71875, + "epoch": 0.2890376035691523, + "grad_norm": 11.139300346374512, + "kl": 0.158203125, + "learning_rate": 7.109623964308477e-07, + "loss": 0.0063, + "reward": 1.453099012374878, + "reward_std": 0.09419392049312592, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4530990421772003, + "step": 907 + }, + { + "completion_length": 121.375, + "epoch": 0.28935627788400253, + "grad_norm": 22.76377296447754, + "kl": 0.1328125, + "learning_rate": 7.106437221159975e-07, + "loss": 0.0053, + "reward": 1.3690372705459595, + "reward_std": 0.08764107525348663, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.36903730034828186, + "rewards/pad": 0.0, + "step": 908 + }, + { + "completion_length": 97.6875, + "epoch": 0.28967495219885275, + "grad_norm": 36.15110397338867, + "kl": 0.1474609375, + "learning_rate": 7.103250478011472e-07, + "loss": 0.0059, + "reward": 1.514521837234497, + "reward_std": 0.09117848426103592, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5145218372344971, + "rewards/pad": 0.0, + "step": 909 + }, + { + "completion_length": 45.015625, + "epoch": 0.289993626513703, + "grad_norm": 43.60408020019531, + "kl": 0.212890625, + "learning_rate": 7.10006373486297e-07, + "loss": 0.0085, + "reward": 1.572979211807251, + "reward_std": 0.15550121665000916, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5729792714118958, + "rewards/pad": 0.0, + "step": 910 + }, + { + "completion_length": 123.15625, + "epoch": 0.2903123008285532, + "grad_norm": 5.848013877868652, + "kl": 0.1396484375, + "learning_rate": 7.096876991714468e-07, + "loss": 0.0056, + "reward": 1.5237162113189697, + "reward_std": 0.10056842863559723, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5237162113189697, + "step": 911 + }, + { + "completion_length": 74.34375, + "epoch": 0.29063097514340347, + "grad_norm": 28.616756439208984, + "kl": 0.1142578125, + "learning_rate": 7.093690248565966e-07, + "loss": 0.0046, + "reward": 1.6580133438110352, + "reward_std": 0.21618688106536865, + "rewards/pad": 0.21875, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.43926331400871277, + "step": 912 + }, + { + "completion_length": 71.71875, + "epoch": 0.2909496494582537, + "grad_norm": 10.530069351196289, + "kl": 0.1123046875, + "learning_rate": 7.090503505417463e-07, + "loss": 0.0045, + "reward": 1.6587841510772705, + "reward_std": 0.14223003387451172, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.5494092702865601, + "rewards/pad": 0.125, + "step": 913 + }, + { + "completion_length": 128.546875, + "epoch": 0.2912683237731039, + "grad_norm": 24.983131408691406, + "kl": 0.1396484375, + "learning_rate": 7.087316762268961e-07, + "loss": 0.0056, + "reward": 1.6001564264297485, + "reward_std": 0.08532114326953888, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.35015642642974854, + "rewards/pad": 0.25, + "step": 914 + }, + { + "completion_length": 154.4375, + "epoch": 0.29158699808795413, + "grad_norm": 12.255154609680176, + "kl": 0.1015625, + "learning_rate": 7.084130019120459e-07, + "loss": 0.0041, + "reward": 1.4581921100616455, + "reward_std": 0.10803060233592987, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.45819202065467834, + "step": 915 + }, + { + "completion_length": 96.484375, + "epoch": 0.29190567240280435, + "grad_norm": 16.626344680786133, + "kl": 0.2734375, + "learning_rate": 7.080943275971957e-07, + "loss": 0.011, + "reward": 1.5853779315948486, + "reward_std": 0.09588365256786346, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5853779911994934, + "rewards/pad": 0.0, + "step": 916 + }, + { + "completion_length": 72.53125, + "epoch": 0.29222434671765457, + "grad_norm": 100.90467071533203, + "kl": 0.0986328125, + "learning_rate": 7.077756532823454e-07, + "loss": 0.004, + "reward": 1.6992861032485962, + "reward_std": 0.10677085071802139, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5742862224578857, + "rewards/pad": 0.125, + "step": 917 + }, + { + "completion_length": 72.609375, + "epoch": 0.2925430210325048, + "grad_norm": 55.456092834472656, + "kl": 0.1630859375, + "learning_rate": 7.074569789674952e-07, + "loss": 0.0065, + "reward": 1.8140954971313477, + "reward_std": 0.1802692860364914, + "rewards/answer_reward": 0.265625, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5484704971313477, + "step": 918 + }, + { + "completion_length": 126.328125, + "epoch": 0.292861695347355, + "grad_norm": 63.22246170043945, + "kl": 0.09521484375, + "learning_rate": 7.07138304652645e-07, + "loss": 0.0038, + "reward": 1.5442649126052856, + "reward_std": 0.05458119511604309, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4192649722099304, + "step": 919 + }, + { + "completion_length": 82.546875, + "epoch": 0.29318036966220523, + "grad_norm": 33.70513916015625, + "kl": 0.1357421875, + "learning_rate": 7.068196303377948e-07, + "loss": 0.0054, + "reward": 1.5909119844436646, + "reward_std": 0.13808241486549377, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.46591198444366455, + "rewards/pad": 0.125, + "step": 920 + }, + { + "completion_length": 97.546875, + "epoch": 0.29349904397705545, + "grad_norm": 22.9251651763916, + "kl": 0.12060546875, + "learning_rate": 7.065009560229445e-07, + "loss": 0.0048, + "reward": 1.4877631664276123, + "reward_std": 0.07348068058490753, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.48776310682296753, + "step": 921 + }, + { + "completion_length": 71.5625, + "epoch": 0.29381771829190567, + "grad_norm": 41.56142044067383, + "kl": 0.1416015625, + "learning_rate": 7.061822817080942e-07, + "loss": 0.0057, + "reward": 1.4704723358154297, + "reward_std": 0.18349112570285797, + "rewards/pad": 0.03125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4392223358154297, + "step": 922 + }, + { + "completion_length": 72.96875, + "epoch": 0.2941363926067559, + "grad_norm": 23.7952823638916, + "kl": 0.1728515625, + "learning_rate": 7.05863607393244e-07, + "loss": 0.0069, + "reward": 1.7845497131347656, + "reward_std": 0.07781291007995605, + "rewards/pad": 0.375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4095497131347656, + "step": 923 + }, + { + "completion_length": 20.125, + "epoch": 0.2944550669216061, + "grad_norm": 30.204940795898438, + "kl": 0.15625, + "learning_rate": 7.055449330783938e-07, + "loss": 0.0062, + "reward": 1.6023484468460083, + "reward_std": 0.14261382818222046, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3992233872413635, + "rewards/pad": 0.203125, + "step": 924 + }, + { + "completion_length": 99.046875, + "epoch": 0.29477374123645633, + "grad_norm": 23.71004867553711, + "kl": 0.087890625, + "learning_rate": 7.052262587635436e-07, + "loss": 0.0035, + "reward": 1.5678212642669678, + "reward_std": 0.07591604441404343, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.44282129406929016, + "rewards/pad": 0.125, + "step": 925 + }, + { + "completion_length": 173.90625, + "epoch": 0.29509241555130655, + "grad_norm": 24.144794464111328, + "kl": 0.076171875, + "learning_rate": 7.049075844486934e-07, + "loss": 0.0031, + "reward": 1.3976125717163086, + "reward_std": 0.08255143463611603, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3976125717163086, + "rewards/pad": 0.0, + "step": 926 + }, + { + "completion_length": 98.3125, + "epoch": 0.2954110898661568, + "grad_norm": 18.583818435668945, + "kl": 0.10546875, + "learning_rate": 7.045889101338432e-07, + "loss": 0.0042, + "reward": 1.6382040977478027, + "reward_std": 0.12732411921024323, + "rewards/pad": 0.09375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.544454038143158, + "step": 927 + }, + { + "completion_length": 70.078125, + "epoch": 0.295729764181007, + "grad_norm": 19.732210159301758, + "kl": 0.1806640625, + "learning_rate": 7.042702358189929e-07, + "loss": 0.0072, + "reward": 1.6882609128952026, + "reward_std": 0.10261107981204987, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5632609128952026, + "rewards/pad": 0.125, + "step": 928 + }, + { + "completion_length": 68.578125, + "epoch": 0.2960484384958572, + "grad_norm": 21.04136085510254, + "kl": 0.173828125, + "learning_rate": 7.039515615041427e-07, + "loss": 0.0069, + "reward": 1.5921133756637573, + "reward_std": 0.1815553903579712, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4983634352684021, + "rewards/pad": 0.09375, + "step": 929 + }, + { + "completion_length": 71.6875, + "epoch": 0.29636711281070743, + "grad_norm": 30.88709259033203, + "kl": 0.1865234375, + "learning_rate": 7.036328871892925e-07, + "loss": 0.0075, + "reward": 1.7662714719772339, + "reward_std": 0.10816091299057007, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5162714123725891, + "rewards/pad": 0.25, + "step": 930 + }, + { + "completion_length": 148.5, + "epoch": 0.29668578712555765, + "grad_norm": 8.878599166870117, + "kl": 0.06884765625, + "learning_rate": 7.033142128744423e-07, + "loss": 0.0028, + "reward": 1.5028890371322632, + "reward_std": 0.07813405245542526, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.37788909673690796, + "step": 931 + }, + { + "completion_length": 97.828125, + "epoch": 0.29700446144040793, + "grad_norm": 50.868247985839844, + "kl": 0.10888671875, + "learning_rate": 7.02995538559592e-07, + "loss": 0.0044, + "reward": 1.5021045207977295, + "reward_std": 0.10513219237327576, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3771045506000519, + "rewards/pad": 0.125, + "step": 932 + }, + { + "completion_length": 70.15625, + "epoch": 0.29732313575525815, + "grad_norm": 16.90110206604004, + "kl": 0.12451171875, + "learning_rate": 7.026768642447418e-07, + "loss": 0.005, + "reward": 1.599424958229065, + "reward_std": 0.06715093553066254, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5994249582290649, + "step": 933 + }, + { + "completion_length": 100.515625, + "epoch": 0.29764181007010837, + "grad_norm": 18.996936798095703, + "kl": 0.1396484375, + "learning_rate": 7.023581899298916e-07, + "loss": 0.0056, + "reward": 1.7346221208572388, + "reward_std": 0.12144883722066879, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.45337218046188354, + "rewards/pad": 0.28125, + "step": 934 + }, + { + "completion_length": 122.171875, + "epoch": 0.2979604843849586, + "grad_norm": 36.10796356201172, + "kl": 0.08056640625, + "learning_rate": 7.020395156150414e-07, + "loss": 0.0032, + "reward": 1.5318636894226074, + "reward_std": 0.05889037996530533, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.40686365962028503, + "step": 935 + }, + { + "completion_length": 200.625, + "epoch": 0.2982791586998088, + "grad_norm": 10.24109172821045, + "kl": 0.05859375, + "learning_rate": 7.017208413001911e-07, + "loss": 0.0023, + "reward": 1.5260357856750488, + "reward_std": 0.05816100537776947, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5260357856750488, + "step": 936 + }, + { + "completion_length": 71.875, + "epoch": 0.29859783301465903, + "grad_norm": 20.65817642211914, + "kl": 0.10498046875, + "learning_rate": 7.014021669853409e-07, + "loss": 0.0042, + "reward": 1.8145759105682373, + "reward_std": 0.04877658560872078, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6895760297775269, + "rewards/pad": 0.125, + "step": 937 + }, + { + "completion_length": 96.765625, + "epoch": 0.29891650732950925, + "grad_norm": 63.60162353515625, + "kl": 0.09375, + "learning_rate": 7.010834926704907e-07, + "loss": 0.0038, + "reward": 1.662316083908081, + "reward_std": 0.1494637280702591, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.5529410243034363, + "rewards/pad": 0.125, + "step": 938 + }, + { + "completion_length": 73.0, + "epoch": 0.29923518164435947, + "grad_norm": 20.451332092285156, + "kl": 0.1787109375, + "learning_rate": 7.007648183556405e-07, + "loss": 0.0072, + "reward": 1.5908095836639404, + "reward_std": 0.10362397134304047, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.46580955386161804, + "step": 939 + }, + { + "completion_length": 98.984375, + "epoch": 0.2995538559592097, + "grad_norm": 76.37430572509766, + "kl": 0.099609375, + "learning_rate": 7.004461440407902e-07, + "loss": 0.004, + "reward": 1.584186315536499, + "reward_std": 0.05407923460006714, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4591863751411438, + "step": 940 + }, + { + "completion_length": 150.171875, + "epoch": 0.2998725302740599, + "grad_norm": 22.983171463012695, + "kl": 0.087890625, + "learning_rate": 7.0012746972594e-07, + "loss": 0.0035, + "reward": 1.4531705379486084, + "reward_std": 0.07982495427131653, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.45317062735557556, + "step": 941 + }, + { + "completion_length": 68.921875, + "epoch": 0.30019120458891013, + "grad_norm": 33.193077087402344, + "kl": 0.1435546875, + "learning_rate": 6.998087954110899e-07, + "loss": 0.0057, + "reward": 1.7539644241333008, + "reward_std": 0.18200719356536865, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.45708948373794556, + "rewards/pad": 0.296875, + "step": 942 + }, + { + "completion_length": 96.5, + "epoch": 0.30050987890376035, + "grad_norm": 16.90999412536621, + "kl": 0.16015625, + "learning_rate": 6.994901210962397e-07, + "loss": 0.0064, + "reward": 1.6284356117248535, + "reward_std": 0.10825789719820023, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5034356117248535, + "rewards/pad": 0.125, + "step": 943 + }, + { + "completion_length": 44.109375, + "epoch": 0.3008285532186106, + "grad_norm": 42.69044494628906, + "kl": 0.1796875, + "learning_rate": 6.991714467813894e-07, + "loss": 0.0072, + "reward": 1.3881826400756836, + "reward_std": 0.10093092918395996, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.2631826102733612, + "rewards/pad": 0.125, + "step": 944 + }, + { + "completion_length": 97.296875, + "epoch": 0.3011472275334608, + "grad_norm": 51.676631927490234, + "kl": 0.267578125, + "learning_rate": 6.988527724665392e-07, + "loss": 0.0107, + "reward": 1.3762681484222412, + "reward_std": 0.11830200254917145, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3762681484222412, + "rewards/pad": 0.0, + "step": 945 + }, + { + "completion_length": 124.5, + "epoch": 0.301465901848311, + "grad_norm": 36.757896423339844, + "kl": 0.06591796875, + "learning_rate": 6.98534098151689e-07, + "loss": 0.0026, + "reward": 1.6907156705856323, + "reward_std": 0.12420570105314255, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5344657301902771, + "rewards/pad": 0.15625, + "step": 946 + }, + { + "completion_length": 71.96875, + "epoch": 0.30178457616316123, + "grad_norm": 60.29111099243164, + "kl": 0.1435546875, + "learning_rate": 6.982154238368388e-07, + "loss": 0.0057, + "reward": 1.5995805263519287, + "reward_std": 0.11470350623130798, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5995805859565735, + "rewards/pad": 0.0, + "step": 947 + }, + { + "completion_length": 147.796875, + "epoch": 0.30210325047801145, + "grad_norm": 10.41659927368164, + "kl": 0.07666015625, + "learning_rate": 6.978967495219885e-07, + "loss": 0.0031, + "reward": 1.614039421081543, + "reward_std": 0.06173337250947952, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.48903945088386536, + "step": 948 + }, + { + "completion_length": 72.875, + "epoch": 0.3024219247928617, + "grad_norm": 34.23875045776367, + "kl": 0.140625, + "learning_rate": 6.975780752071383e-07, + "loss": 0.0056, + "reward": 1.671351671218872, + "reward_std": 0.11177132278680801, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6244765520095825, + "rewards/pad": 0.046875, + "step": 949 + }, + { + "completion_length": 95.75, + "epoch": 0.3027405991077119, + "grad_norm": 45.31456756591797, + "kl": 0.1298828125, + "learning_rate": 6.972594008922881e-07, + "loss": 0.0052, + "reward": 1.6682624816894531, + "reward_std": 0.12410464882850647, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5432624816894531, + "rewards/pad": 0.125, + "step": 950 + }, + { + "completion_length": 121.0625, + "epoch": 0.3030592734225621, + "grad_norm": 29.192468643188477, + "kl": 0.10888671875, + "learning_rate": 6.969407265774379e-07, + "loss": 0.0044, + "reward": 1.4336252212524414, + "reward_std": 0.06404151022434235, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.43362507224082947, + "step": 951 + }, + { + "completion_length": 174.796875, + "epoch": 0.3033779477374124, + "grad_norm": 7.179665565490723, + "kl": 0.1259765625, + "learning_rate": 6.966220522625876e-07, + "loss": 0.005, + "reward": 1.3703272342681885, + "reward_std": 0.047330766916275024, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.37032726407051086, + "step": 952 + }, + { + "completion_length": 70.0625, + "epoch": 0.3036966220522626, + "grad_norm": 17.197315216064453, + "kl": 0.150390625, + "learning_rate": 6.963033779477374e-07, + "loss": 0.006, + "reward": 1.5124963521957397, + "reward_std": 0.10813236236572266, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5124963521957397, + "rewards/pad": 0.0, + "step": 953 + }, + { + "completion_length": 121.34375, + "epoch": 0.30401529636711283, + "grad_norm": 38.938541412353516, + "kl": 0.09033203125, + "learning_rate": 6.959847036328872e-07, + "loss": 0.0036, + "reward": 1.638798713684082, + "reward_std": 0.05554702877998352, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5137988924980164, + "step": 954 + }, + { + "completion_length": 122.359375, + "epoch": 0.30433397068196305, + "grad_norm": 21.31637954711914, + "kl": 0.1083984375, + "learning_rate": 6.95666029318037e-07, + "loss": 0.0043, + "reward": 1.507123351097107, + "reward_std": 0.07658617943525314, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5071233510971069, + "step": 955 + }, + { + "completion_length": 171.71875, + "epoch": 0.30465264499681327, + "grad_norm": 25.84273338317871, + "kl": 0.07666015625, + "learning_rate": 6.953473550031867e-07, + "loss": 0.0031, + "reward": 1.4941716194152832, + "reward_std": 0.06094829738140106, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.494171679019928, + "step": 956 + }, + { + "completion_length": 96.71875, + "epoch": 0.3049713193116635, + "grad_norm": 17.762001037597656, + "kl": 0.09375, + "learning_rate": 6.950286806883365e-07, + "loss": 0.0037, + "reward": 1.5891180038452148, + "reward_std": 0.043381866067647934, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5891180038452148, + "rewards/pad": 0.0, + "step": 957 + }, + { + "completion_length": 71.90625, + "epoch": 0.3052899936265137, + "grad_norm": 16.231908798217773, + "kl": 0.12255859375, + "learning_rate": 6.947100063734863e-07, + "loss": 0.0049, + "reward": 1.5764062404632568, + "reward_std": 0.06593342125415802, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.32640618085861206, + "rewards/pad": 0.25, + "step": 958 + }, + { + "completion_length": 96.015625, + "epoch": 0.30560866794136393, + "grad_norm": 62.19764709472656, + "kl": 0.091796875, + "learning_rate": 6.94391332058636e-07, + "loss": 0.0037, + "reward": 1.4593048095703125, + "reward_std": 0.06122640520334244, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3343048095703125, + "step": 959 + }, + { + "completion_length": 72.53125, + "epoch": 0.30592734225621415, + "grad_norm": 25.66366195678711, + "kl": 0.185546875, + "learning_rate": 6.940726577437858e-07, + "loss": 0.0074, + "reward": 1.6957283020019531, + "reward_std": 0.08966457843780518, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5707283020019531, + "rewards/pad": 0.125, + "step": 960 + }, + { + "completion_length": 95.5, + "epoch": 0.3062460165710644, + "grad_norm": 23.313758850097656, + "kl": 0.212890625, + "learning_rate": 6.937539834289357e-07, + "loss": 0.0085, + "reward": 1.4860730171203613, + "reward_std": 0.097783163189888, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4860730767250061, + "step": 961 + }, + { + "completion_length": 45.953125, + "epoch": 0.3065646908859146, + "grad_norm": 27.550352096557617, + "kl": 0.2138671875, + "learning_rate": 6.934353091140854e-07, + "loss": 0.0086, + "reward": 1.809466004371643, + "reward_std": 0.11983469128608704, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5750910043716431, + "rewards/pad": 0.234375, + "step": 962 + }, + { + "completion_length": 44.515625, + "epoch": 0.3068833652007648, + "grad_norm": 86.11719512939453, + "kl": 0.2216796875, + "learning_rate": 6.931166347992351e-07, + "loss": 0.0089, + "reward": 1.5869412422180176, + "reward_std": 0.1016860380768776, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5869413018226624, + "rewards/pad": 0.0, + "step": 963 + }, + { + "completion_length": 173.953125, + "epoch": 0.30720203951561503, + "grad_norm": 15.720704078674316, + "kl": 0.103515625, + "learning_rate": 6.927979604843849e-07, + "loss": 0.0041, + "reward": 1.4632954597473145, + "reward_std": 0.04079699516296387, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.46329545974731445, + "step": 964 + }, + { + "completion_length": 96.53125, + "epoch": 0.30752071383046525, + "grad_norm": 21.17584800720215, + "kl": 0.1357421875, + "learning_rate": 6.924792861695347e-07, + "loss": 0.0054, + "reward": 1.526242733001709, + "reward_std": 0.10939948260784149, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.49499279260635376, + "rewards/pad": 0.03125, + "step": 965 + }, + { + "completion_length": 97.109375, + "epoch": 0.3078393881453155, + "grad_norm": 19.60423469543457, + "kl": 0.1337890625, + "learning_rate": 6.921606118546845e-07, + "loss": 0.0053, + "reward": 1.5197116136550903, + "reward_std": 0.05108891427516937, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5197116136550903, + "step": 966 + }, + { + "completion_length": 70.5, + "epoch": 0.3081580624601657, + "grad_norm": 62.79570007324219, + "kl": 0.16796875, + "learning_rate": 6.918419375398342e-07, + "loss": 0.0067, + "reward": 1.4708547592163086, + "reward_std": 0.06340339034795761, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.34585481882095337, + "step": 967 + }, + { + "completion_length": 96.34375, + "epoch": 0.3084767367750159, + "grad_norm": 24.94214630126953, + "kl": 0.162109375, + "learning_rate": 6.91523263224984e-07, + "loss": 0.0065, + "reward": 1.4744987487792969, + "reward_std": 0.06317903846502304, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4744986891746521, + "rewards/pad": 0.0, + "step": 968 + }, + { + "completion_length": 43.8125, + "epoch": 0.30879541108986613, + "grad_norm": 30.116613388061523, + "kl": 0.1904296875, + "learning_rate": 6.912045889101338e-07, + "loss": 0.0076, + "reward": 1.6158849000930786, + "reward_std": 0.07592947781085968, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4908849000930786, + "rewards/pad": 0.125, + "step": 969 + }, + { + "completion_length": 71.796875, + "epoch": 0.30911408540471635, + "grad_norm": 130.30897521972656, + "kl": 0.193359375, + "learning_rate": 6.908859145952836e-07, + "loss": 0.0078, + "reward": 1.4907974004745483, + "reward_std": 0.11981174349784851, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.49079734086990356, + "rewards/pad": 0.0, + "step": 970 + }, + { + "completion_length": 71.609375, + "epoch": 0.30943275971956663, + "grad_norm": 73.44354248046875, + "kl": 0.134765625, + "learning_rate": 6.905672402804333e-07, + "loss": 0.0054, + "reward": 1.5406136512756348, + "reward_std": 0.05049721896648407, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5406137704849243, + "rewards/pad": 0.0, + "step": 971 + }, + { + "completion_length": 69.96875, + "epoch": 0.30975143403441685, + "grad_norm": 21.05704116821289, + "kl": 0.166015625, + "learning_rate": 6.902485659655831e-07, + "loss": 0.0066, + "reward": 1.5726382732391357, + "reward_std": 0.08547885715961456, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5726381540298462, + "rewards/pad": 0.0, + "step": 972 + }, + { + "completion_length": 144.015625, + "epoch": 0.31007010834926707, + "grad_norm": 36.80889129638672, + "kl": 0.0791015625, + "learning_rate": 6.899298916507329e-07, + "loss": 0.0031, + "reward": 1.6074941158294678, + "reward_std": 0.09198617935180664, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6074941158294678, + "rewards/pad": 0.0, + "step": 973 + }, + { + "completion_length": 123.59375, + "epoch": 0.3103887826641173, + "grad_norm": 13.655397415161133, + "kl": 0.11767578125, + "learning_rate": 6.896112173358827e-07, + "loss": 0.0047, + "reward": 1.6335394382476807, + "reward_std": 0.06385113298892975, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5085393190383911, + "step": 974 + }, + { + "completion_length": 122.453125, + "epoch": 0.3107074569789675, + "grad_norm": 113.74642944335938, + "kl": 0.1259765625, + "learning_rate": 6.892925430210324e-07, + "loss": 0.005, + "reward": 1.5529446601867676, + "reward_std": 0.112543486058712, + "rewards/pad": 0.0625, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4904446005821228, + "step": 975 + }, + { + "completion_length": 121.265625, + "epoch": 0.31102613129381773, + "grad_norm": 29.511276245117188, + "kl": 0.1279296875, + "learning_rate": 6.889738687061822e-07, + "loss": 0.0051, + "reward": 1.3846508264541626, + "reward_std": 0.03916772082448006, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3846507966518402, + "step": 976 + }, + { + "completion_length": 46.671875, + "epoch": 0.31134480560866795, + "grad_norm": 40.682498931884766, + "kl": 0.14453125, + "learning_rate": 6.88655194391332e-07, + "loss": 0.0058, + "reward": 1.8048152923583984, + "reward_std": 0.0671246275305748, + "rewards/answer_reward": 0.375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4298152029514313, + "step": 977 + }, + { + "completion_length": 94.640625, + "epoch": 0.31166347992351817, + "grad_norm": 15.610057830810547, + "kl": 0.12060546875, + "learning_rate": 6.883365200764818e-07, + "loss": 0.0048, + "reward": 1.5543341636657715, + "reward_std": 0.06744150072336197, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5543341636657715, + "rewards/pad": 0.0, + "step": 978 + }, + { + "completion_length": 145.6875, + "epoch": 0.3119821542383684, + "grad_norm": 34.307952880859375, + "kl": 0.11279296875, + "learning_rate": 6.880178457616315e-07, + "loss": 0.0045, + "reward": 1.4717872142791748, + "reward_std": 0.11070170998573303, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.4874122738838196, + "step": 979 + }, + { + "completion_length": 98.890625, + "epoch": 0.3123008285532186, + "grad_norm": 74.25923919677734, + "kl": 0.11083984375, + "learning_rate": 6.876991714467814e-07, + "loss": 0.0044, + "reward": 1.5795928239822388, + "reward_std": 0.06632921099662781, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.32959282398223877, + "step": 980 + }, + { + "completion_length": 148.171875, + "epoch": 0.31261950286806883, + "grad_norm": 14.847012519836426, + "kl": 0.1220703125, + "learning_rate": 6.873804971319312e-07, + "loss": 0.0049, + "reward": 1.5648550987243652, + "reward_std": 0.05563882738351822, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.43985506892204285, + "step": 981 + }, + { + "completion_length": 149.90625, + "epoch": 0.31293817718291905, + "grad_norm": 21.062965393066406, + "kl": 0.07861328125, + "learning_rate": 6.87061822817081e-07, + "loss": 0.0031, + "reward": 1.6611219644546509, + "reward_std": 0.04433588683605194, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5361219644546509, + "step": 982 + }, + { + "completion_length": 121.296875, + "epoch": 0.3132568514977693, + "grad_norm": 8.826406478881836, + "kl": 0.08544921875, + "learning_rate": 6.867431485022307e-07, + "loss": 0.0034, + "reward": 1.6873703002929688, + "reward_std": 0.04004920646548271, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4373703598976135, + "rewards/pad": 0.25, + "step": 983 + }, + { + "completion_length": 70.59375, + "epoch": 0.3135755258126195, + "grad_norm": 38.75087356567383, + "kl": 0.1826171875, + "learning_rate": 6.864244741873805e-07, + "loss": 0.0073, + "reward": 1.514086127281189, + "reward_std": 0.13811108469963074, + "rewards/answer_reward": 0.15625, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.3578360974788666, + "step": 984 + }, + { + "completion_length": 121.078125, + "epoch": 0.3138942001274697, + "grad_norm": 23.378746032714844, + "kl": 0.14453125, + "learning_rate": 6.861057998725303e-07, + "loss": 0.0058, + "reward": 1.6368203163146973, + "reward_std": 0.11664395034313202, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5118203163146973, + "step": 985 + }, + { + "completion_length": 97.3125, + "epoch": 0.31421287444231993, + "grad_norm": 65.3720703125, + "kl": 0.1103515625, + "learning_rate": 6.8578712555768e-07, + "loss": 0.0044, + "reward": 1.5769407749176025, + "reward_std": 0.1618584394454956, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4519408047199249, + "step": 986 + }, + { + "completion_length": 121.75, + "epoch": 0.31453154875717015, + "grad_norm": 53.989322662353516, + "kl": 0.1552734375, + "learning_rate": 6.854684512428298e-07, + "loss": 0.0062, + "reward": 1.5589345693588257, + "reward_std": 0.05939555913209915, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5589345693588257, + "rewards/pad": 0.0, + "step": 987 + }, + { + "completion_length": 171.234375, + "epoch": 0.3148502230720204, + "grad_norm": 12.67489242553711, + "kl": 0.06689453125, + "learning_rate": 6.851497769279796e-07, + "loss": 0.0027, + "reward": 1.4538019895553589, + "reward_std": 0.03941020369529724, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4538019597530365, + "rewards/pad": 0.0, + "step": 988 + }, + { + "completion_length": 120.84375, + "epoch": 0.3151688973868706, + "grad_norm": 21.424421310424805, + "kl": 0.12451171875, + "learning_rate": 6.848311026131294e-07, + "loss": 0.005, + "reward": 1.5307704210281372, + "reward_std": 0.1141161248087883, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.546395480632782, + "step": 989 + }, + { + "completion_length": 96.46875, + "epoch": 0.3154875717017208, + "grad_norm": 28.962202072143555, + "kl": 0.1103515625, + "learning_rate": 6.845124282982791e-07, + "loss": 0.0044, + "reward": 1.519028902053833, + "reward_std": 0.04207426309585571, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.39402878284454346, + "step": 990 + }, + { + "completion_length": 97.796875, + "epoch": 0.3158062460165711, + "grad_norm": 126.08263397216797, + "kl": 0.11376953125, + "learning_rate": 6.841937539834289e-07, + "loss": 0.0046, + "reward": 1.5184412002563477, + "reward_std": 0.06963367760181427, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.2684412896633148, + "rewards/pad": 0.25, + "step": 991 + }, + { + "completion_length": 67.875, + "epoch": 0.3161249203314213, + "grad_norm": 66.68081665039062, + "kl": 0.1748046875, + "learning_rate": 6.838750796685787e-07, + "loss": 0.007, + "reward": 1.6175551414489746, + "reward_std": 0.07942277193069458, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.6175551414489746, + "step": 992 + }, + { + "completion_length": 97.359375, + "epoch": 0.31644359464627153, + "grad_norm": 23.92911148071289, + "kl": 0.255859375, + "learning_rate": 6.835564053537285e-07, + "loss": 0.0102, + "reward": 1.238165020942688, + "reward_std": 0.08866700530052185, + "rewards/pad": 0.015625, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.222540020942688, + "step": 993 + }, + { + "completion_length": 122.703125, + "epoch": 0.31676226896112175, + "grad_norm": 97.40007019042969, + "kl": 0.08544921875, + "learning_rate": 6.832377310388782e-07, + "loss": 0.0034, + "reward": 1.5865464210510254, + "reward_std": 0.18776553869247437, + "rewards/pad": 0.265625, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.32092148065567017, + "step": 994 + }, + { + "completion_length": 119.75, + "epoch": 0.31708094327597197, + "grad_norm": 50.509674072265625, + "kl": 0.125, + "learning_rate": 6.82919056724028e-07, + "loss": 0.005, + "reward": 1.5579121112823486, + "reward_std": 0.04533017426729202, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.43291202187538147, + "rewards/pad": 0.125, + "step": 995 + }, + { + "completion_length": 173.40625, + "epoch": 0.3173996175908222, + "grad_norm": 9.46157169342041, + "kl": 0.07958984375, + "learning_rate": 6.826003824091778e-07, + "loss": 0.0032, + "reward": 1.315908432006836, + "reward_std": 0.02767709270119667, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3159083127975464, + "step": 996 + }, + { + "completion_length": 95.578125, + "epoch": 0.3177182919056724, + "grad_norm": 181.33319091796875, + "kl": 0.1455078125, + "learning_rate": 6.822817080943276e-07, + "loss": 0.0058, + "reward": 1.6179592609405518, + "reward_std": 0.1326274871826172, + "rewards/answer_reward": 0.09375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5242092609405518, + "step": 997 + }, + { + "completion_length": 70.15625, + "epoch": 0.31803696622052263, + "grad_norm": 24.98843765258789, + "kl": 0.26953125, + "learning_rate": 6.819630337794773e-07, + "loss": 0.0108, + "reward": 1.6061959266662598, + "reward_std": 0.07357417047023773, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.606195867061615, + "step": 998 + }, + { + "completion_length": 70.953125, + "epoch": 0.31835564053537285, + "grad_norm": 21.293861389160156, + "kl": 0.2080078125, + "learning_rate": 6.816443594646272e-07, + "loss": 0.0083, + "reward": 1.7381505966186523, + "reward_std": 0.07790133357048035, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6131504774093628, + "rewards/pad": 0.125, + "step": 999 + }, + { + "completion_length": 123.15625, + "epoch": 0.3186743148502231, + "grad_norm": 103.16876220703125, + "kl": 0.107421875, + "learning_rate": 6.81325685149777e-07, + "loss": 0.0043, + "reward": 1.484614372253418, + "reward_std": 0.07241792976856232, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3596143126487732, + "step": 1000 + }, + { + "completion_length": 149.5625, + "epoch": 0.3189929891650733, + "grad_norm": 9.273104667663574, + "kl": 0.08056640625, + "learning_rate": 6.810070108349267e-07, + "loss": 0.0032, + "reward": 1.6335794925689697, + "reward_std": 0.03922747075557709, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3835795819759369, + "step": 1001 + }, + { + "completion_length": 120.96875, + "epoch": 0.3193116634799235, + "grad_norm": 27.187664031982422, + "kl": 0.166015625, + "learning_rate": 6.806883365200764e-07, + "loss": 0.0066, + "reward": 1.4678020477294922, + "reward_std": 0.04097019135951996, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4678019881248474, + "rewards/pad": 0.0, + "step": 1002 + }, + { + "completion_length": 123.53125, + "epoch": 0.31963033779477373, + "grad_norm": 6.377375602722168, + "kl": 0.12060546875, + "learning_rate": 6.803696622052262e-07, + "loss": 0.0048, + "reward": 1.4584826231002808, + "reward_std": 0.03448522090911865, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.45848262310028076, + "step": 1003 + }, + { + "completion_length": 43.953125, + "epoch": 0.31994901210962395, + "grad_norm": 53.49687576293945, + "kl": 0.181640625, + "learning_rate": 6.80050987890376e-07, + "loss": 0.0073, + "reward": 1.544199824333191, + "reward_std": 0.04130152612924576, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5441998243331909, + "rewards/pad": 0.0, + "step": 1004 + }, + { + "completion_length": 20.203125, + "epoch": 0.3202676864244742, + "grad_norm": 69.30970001220703, + "kl": 0.1748046875, + "learning_rate": 6.797323135755258e-07, + "loss": 0.007, + "reward": 1.620455265045166, + "reward_std": 0.12522095441818237, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.49545520544052124, + "rewards/pad": 0.125, + "step": 1005 + }, + { + "completion_length": 42.375, + "epoch": 0.3205863607393244, + "grad_norm": 104.5580825805664, + "kl": 0.185546875, + "learning_rate": 6.794136392606755e-07, + "loss": 0.0074, + "reward": 1.6227277517318726, + "reward_std": 0.10305334627628326, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5289777517318726, + "rewards/pad": 0.09375, + "step": 1006 + }, + { + "completion_length": 121.515625, + "epoch": 0.3209050350541746, + "grad_norm": 82.89877319335938, + "kl": 0.12890625, + "learning_rate": 6.790949649458253e-07, + "loss": 0.0051, + "reward": 1.4882017374038696, + "reward_std": 0.041941650211811066, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.48820173740386963, + "step": 1007 + }, + { + "completion_length": 95.71875, + "epoch": 0.32122370936902483, + "grad_norm": 19.741044998168945, + "kl": 0.10986328125, + "learning_rate": 6.787762906309751e-07, + "loss": 0.0044, + "reward": 1.619329810142517, + "reward_std": 0.04831254482269287, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4943298101425171, + "step": 1008 + }, + { + "completion_length": 118.453125, + "epoch": 0.32154238368387505, + "grad_norm": 42.4965934753418, + "kl": 0.1123046875, + "learning_rate": 6.784576163161249e-07, + "loss": 0.0045, + "reward": 1.4485756158828735, + "reward_std": 0.08220487833023071, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.44857558608055115, + "step": 1009 + }, + { + "completion_length": 48.265625, + "epoch": 0.3218610579987253, + "grad_norm": 32.22761154174805, + "kl": 0.1298828125, + "learning_rate": 6.781389420012746e-07, + "loss": 0.0052, + "reward": 1.849495768547058, + "reward_std": 0.08726021647453308, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.47449585795402527, + "rewards/pad": 0.375, + "step": 1010 + }, + { + "completion_length": 44.859375, + "epoch": 0.32217973231357555, + "grad_norm": 19.2244815826416, + "kl": 0.12060546875, + "learning_rate": 6.778202676864244e-07, + "loss": 0.0048, + "reward": 1.932586908340454, + "reward_std": 0.05833645537495613, + "rewards/answer_reward": 0.375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5575869083404541, + "step": 1011 + }, + { + "completion_length": 144.125, + "epoch": 0.32249840662842577, + "grad_norm": 37.333648681640625, + "kl": 0.10009765625, + "learning_rate": 6.775015933715742e-07, + "loss": 0.004, + "reward": 1.5504242181777954, + "reward_std": 0.03601760044693947, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5504241585731506, + "step": 1012 + }, + { + "completion_length": 97.203125, + "epoch": 0.322817080943276, + "grad_norm": 47.14528274536133, + "kl": 0.1611328125, + "learning_rate": 6.77182919056724e-07, + "loss": 0.0065, + "reward": 1.682951807975769, + "reward_std": 0.08218291401863098, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.557951807975769, + "step": 1013 + }, + { + "completion_length": 119.859375, + "epoch": 0.3231357552581262, + "grad_norm": 36.78887939453125, + "kl": 0.251953125, + "learning_rate": 6.768642447418737e-07, + "loss": 0.0101, + "reward": 1.429707646369934, + "reward_std": 0.08912655711174011, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4297076463699341, + "step": 1014 + }, + { + "completion_length": 99.5, + "epoch": 0.32345442957297643, + "grad_norm": 63.29979705810547, + "kl": 0.138671875, + "learning_rate": 6.765455704270235e-07, + "loss": 0.0055, + "reward": 1.4366942644119263, + "reward_std": 0.08463046699762344, + "rewards/pad": 0.109375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.32731932401657104, + "step": 1015 + }, + { + "completion_length": 66.671875, + "epoch": 0.32377310388782665, + "grad_norm": 39.79801559448242, + "kl": 0.140625, + "learning_rate": 6.762268961121733e-07, + "loss": 0.0056, + "reward": 1.6116578578948975, + "reward_std": 0.09199018031358719, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.48665785789489746, + "step": 1016 + }, + { + "completion_length": 46.09375, + "epoch": 0.3240917782026769, + "grad_norm": 64.3765869140625, + "kl": 0.220703125, + "learning_rate": 6.75908221797323e-07, + "loss": 0.0088, + "reward": 1.686189889907837, + "reward_std": 0.0763259083032608, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5611898899078369, + "rewards/pad": 0.125, + "step": 1017 + }, + { + "completion_length": 146.265625, + "epoch": 0.3244104525175271, + "grad_norm": 46.217708587646484, + "kl": 0.2041015625, + "learning_rate": 6.755895474824729e-07, + "loss": 0.0082, + "reward": 1.3923192024230957, + "reward_std": 0.09687663614749908, + "rewards/answer_reward": 0.0, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.3923192322254181, + "step": 1018 + }, + { + "completion_length": 121.5, + "epoch": 0.3247291268323773, + "grad_norm": 21.247774124145508, + "kl": 0.1572265625, + "learning_rate": 6.752708731676227e-07, + "loss": 0.0063, + "reward": 1.637831211090088, + "reward_std": 0.05769766867160797, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5128312706947327, + "rewards/pad": 0.125, + "step": 1019 + }, + { + "completion_length": 120.734375, + "epoch": 0.32504780114722753, + "grad_norm": 29.704471588134766, + "kl": 0.10302734375, + "learning_rate": 6.749521988527725e-07, + "loss": 0.0041, + "reward": 1.5674422979354858, + "reward_std": 0.07865164428949356, + "rewards/pad": 0.109375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.45806729793548584, + "step": 1020 + }, + { + "completion_length": 93.671875, + "epoch": 0.32536647546207775, + "grad_norm": 36.188941955566406, + "kl": 0.115234375, + "learning_rate": 6.746335245379222e-07, + "loss": 0.0046, + "reward": 1.6116740703582764, + "reward_std": 0.12697413563728333, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.37729912996292114, + "rewards/pad": 0.25, + "step": 1021 + }, + { + "completion_length": 121.6875, + "epoch": 0.325685149776928, + "grad_norm": 22.342357635498047, + "kl": 0.09521484375, + "learning_rate": 6.74314850223072e-07, + "loss": 0.0038, + "reward": 1.5704026222229004, + "reward_std": 0.07594575732946396, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.33602750301361084, + "step": 1022 + }, + { + "completion_length": 121.3125, + "epoch": 0.3260038240917782, + "grad_norm": 15.507749557495117, + "kl": 0.36328125, + "learning_rate": 6.739961759082218e-07, + "loss": 0.0145, + "reward": 1.6103272438049316, + "reward_std": 0.06380373984575272, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.48532721400260925, + "step": 1023 + }, + { + "completion_length": 43.75, + "epoch": 0.3263224984066284, + "grad_norm": 26.68657684326172, + "kl": 0.1796875, + "learning_rate": 6.736775015933716e-07, + "loss": 0.0072, + "reward": 1.4567975997924805, + "reward_std": 0.04004968702793121, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4567975401878357, + "rewards/pad": 0.0, + "step": 1024 + }, + { + "completion_length": 71.171875, + "epoch": 0.32664117272147863, + "grad_norm": 95.6634292602539, + "kl": 0.177734375, + "learning_rate": 6.733588272785213e-07, + "loss": 0.0071, + "reward": 1.6111572980880737, + "reward_std": 0.11770939826965332, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5486572980880737, + "rewards/pad": 0.0625, + "step": 1025 + }, + { + "completion_length": 120.140625, + "epoch": 0.32695984703632885, + "grad_norm": 131.3987579345703, + "kl": 0.083984375, + "learning_rate": 6.730401529636711e-07, + "loss": 0.0034, + "reward": 1.5982779264450073, + "reward_std": 0.10901899635791779, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.3482779264450073, + "step": 1026 + }, + { + "completion_length": 121.03125, + "epoch": 0.3272785213511791, + "grad_norm": 114.73072814941406, + "kl": 0.11279296875, + "learning_rate": 6.727214786488209e-07, + "loss": 0.0045, + "reward": 1.5256927013397217, + "reward_std": 0.16002598404884338, + "rewards/answer_reward": 0.046875, + "rewards/format_reward_gqa": 0.984375, + "rewards/iou_glue_reward": 0.4944427013397217, + "step": 1027 + }, + { + "completion_length": 71.75, + "epoch": 0.3275971956660293, + "grad_norm": 39.72523880004883, + "kl": 0.1572265625, + "learning_rate": 6.724028043339707e-07, + "loss": 0.0063, + "reward": 1.767228603363037, + "reward_std": 0.04902970418334007, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5172286629676819, + "step": 1028 + }, + { + "completion_length": 94.15625, + "epoch": 0.3279158699808795, + "grad_norm": 28.721555709838867, + "kl": 0.1298828125, + "learning_rate": 6.720841300191204e-07, + "loss": 0.0052, + "reward": 1.5470216274261475, + "reward_std": 0.06241071969270706, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5470216870307922, + "step": 1029 + }, + { + "completion_length": 46.609375, + "epoch": 0.3282345442957298, + "grad_norm": 43.542396545410156, + "kl": 0.13671875, + "learning_rate": 6.717654557042702e-07, + "loss": 0.0055, + "reward": 1.7156548500061035, + "reward_std": 0.0609230101108551, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5906547904014587, + "step": 1030 + }, + { + "completion_length": 70.34375, + "epoch": 0.32855321861058, + "grad_norm": 53.10152816772461, + "kl": 0.111328125, + "learning_rate": 6.7144678138942e-07, + "loss": 0.0045, + "reward": 1.6983790397644043, + "reward_std": 0.051853202283382416, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5733789801597595, + "rewards/pad": 0.125, + "step": 1031 + }, + { + "completion_length": 120.609375, + "epoch": 0.32887189292543023, + "grad_norm": 35.33661651611328, + "kl": 0.1494140625, + "learning_rate": 6.711281070745698e-07, + "loss": 0.006, + "reward": 1.6467417478561401, + "reward_std": 0.08918267488479614, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5217417478561401, + "step": 1032 + }, + { + "completion_length": 92.859375, + "epoch": 0.32919056724028045, + "grad_norm": 81.88345336914062, + "kl": 0.154296875, + "learning_rate": 6.708094327597195e-07, + "loss": 0.0062, + "reward": 1.5557001829147339, + "reward_std": 0.06777728348970413, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4307001531124115, + "rewards/pad": 0.125, + "step": 1033 + }, + { + "completion_length": 68.90625, + "epoch": 0.32950924155513067, + "grad_norm": 18.9570369720459, + "kl": 0.26171875, + "learning_rate": 6.704907584448693e-07, + "loss": 0.0104, + "reward": 1.6313979625701904, + "reward_std": 0.06359601020812988, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6313979625701904, + "rewards/pad": 0.0, + "step": 1034 + }, + { + "completion_length": 96.703125, + "epoch": 0.3298279158699809, + "grad_norm": 69.87822723388672, + "kl": 0.12451171875, + "learning_rate": 6.701720841300191e-07, + "loss": 0.005, + "reward": 1.544966220855713, + "reward_std": 0.10156506299972534, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.49809128046035767, + "rewards/pad": 0.046875, + "step": 1035 + }, + { + "completion_length": 69.53125, + "epoch": 0.3301465901848311, + "grad_norm": 17.39261817932129, + "kl": 0.1484375, + "learning_rate": 6.69853409815169e-07, + "loss": 0.0059, + "reward": 1.9315378665924072, + "reward_std": 0.10341665893793106, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.6815378069877625, + "step": 1036 + }, + { + "completion_length": 46.921875, + "epoch": 0.33046526449968133, + "grad_norm": 32.02791976928711, + "kl": 0.2216796875, + "learning_rate": 6.695347355003187e-07, + "loss": 0.0089, + "reward": 1.7051042318344116, + "reward_std": 0.055431973189115524, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.455104261636734, + "step": 1037 + }, + { + "completion_length": 118.015625, + "epoch": 0.33078393881453155, + "grad_norm": 46.765708923339844, + "kl": 0.0986328125, + "learning_rate": 6.692160611854685e-07, + "loss": 0.0039, + "reward": 1.5392026901245117, + "reward_std": 0.03716112673282623, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5392026901245117, + "step": 1038 + }, + { + "completion_length": 96.625, + "epoch": 0.3311026131293818, + "grad_norm": 33.89827346801758, + "kl": 0.1318359375, + "learning_rate": 6.688973868706183e-07, + "loss": 0.0053, + "reward": 1.6119462251663208, + "reward_std": 0.0470452755689621, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.3619462251663208, + "step": 1039 + }, + { + "completion_length": 145.578125, + "epoch": 0.331421287444232, + "grad_norm": 54.471920013427734, + "kl": 0.0830078125, + "learning_rate": 6.68578712555768e-07, + "loss": 0.0033, + "reward": 1.4640541076660156, + "reward_std": 0.04760254919528961, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4640541672706604, + "step": 1040 + }, + { + "completion_length": 46.3125, + "epoch": 0.3317399617590822, + "grad_norm": 53.303070068359375, + "kl": 0.2470703125, + "learning_rate": 6.682600382409177e-07, + "loss": 0.0099, + "reward": 1.6290390491485596, + "reward_std": 0.14533916115760803, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.3946640193462372, + "rewards/pad": 0.25, + "step": 1041 + }, + { + "completion_length": 71.4375, + "epoch": 0.33205863607393243, + "grad_norm": 20.868520736694336, + "kl": 0.11572265625, + "learning_rate": 6.679413639260675e-07, + "loss": 0.0046, + "reward": 1.7570959329605103, + "reward_std": 0.06300647556781769, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6320960521697998, + "rewards/pad": 0.125, + "step": 1042 + }, + { + "completion_length": 18.71875, + "epoch": 0.33237731038878265, + "grad_norm": 29.75905418395996, + "kl": 0.349609375, + "learning_rate": 6.676226896112173e-07, + "loss": 0.014, + "reward": 1.7678983211517334, + "reward_std": 0.08387524634599686, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6428983211517334, + "rewards/pad": 0.125, + "step": 1043 + }, + { + "completion_length": 95.203125, + "epoch": 0.3326959847036329, + "grad_norm": 34.0615119934082, + "kl": 0.2578125, + "learning_rate": 6.67304015296367e-07, + "loss": 0.0103, + "reward": 1.4088149070739746, + "reward_std": 0.04754612594842911, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4088148772716522, + "step": 1044 + }, + { + "completion_length": 68.96875, + "epoch": 0.3330146590184831, + "grad_norm": 219.807861328125, + "kl": 0.259765625, + "learning_rate": 6.669853409815168e-07, + "loss": 0.0103, + "reward": 1.607092261314392, + "reward_std": 0.09234102070331573, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4820922017097473, + "rewards/pad": 0.125, + "step": 1045 + }, + { + "completion_length": 144.09375, + "epoch": 0.3333333333333333, + "grad_norm": 23.311023712158203, + "kl": 0.1142578125, + "learning_rate": 6.666666666666666e-07, + "loss": 0.0046, + "reward": 1.5305759906768799, + "reward_std": 0.06265944242477417, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5305759906768799, + "rewards/pad": 0.0, + "step": 1046 + }, + { + "completion_length": 43.453125, + "epoch": 0.33365200764818354, + "grad_norm": 39.3546028137207, + "kl": 0.306640625, + "learning_rate": 6.663479923518164e-07, + "loss": 0.0123, + "reward": 1.440364122390747, + "reward_std": 0.15070118010044098, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.45598912239074707, + "rewards/pad": 0.0, + "step": 1047 + }, + { + "completion_length": 19.671875, + "epoch": 0.33397068196303376, + "grad_norm": 70.71843719482422, + "kl": 0.515625, + "learning_rate": 6.660293180369661e-07, + "loss": 0.0206, + "reward": 1.607330083847046, + "reward_std": 0.15620729327201843, + "rewards/answer_reward": 0.015625, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5917050242424011, + "step": 1048 + }, + { + "completion_length": 122.453125, + "epoch": 0.334289356277884, + "grad_norm": 27.474184036254883, + "kl": 0.08056640625, + "learning_rate": 6.657106437221159e-07, + "loss": 0.0032, + "reward": 1.4041144847869873, + "reward_std": 0.10201508551836014, + "rewards/pad": 0.140625, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.27911460399627686, + "step": 1049 + }, + { + "completion_length": 120.40625, + "epoch": 0.33460803059273425, + "grad_norm": 28.913145065307617, + "kl": 0.1025390625, + "learning_rate": 6.653919694072657e-07, + "loss": 0.0041, + "reward": 1.5066144466400146, + "reward_std": 0.08884887397289276, + "rewards/answer_reward": 0.0, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5066144466400146, + "step": 1050 + }, + { + "completion_length": 94.984375, + "epoch": 0.33492670490758447, + "grad_norm": 82.88301086425781, + "kl": 0.1591796875, + "learning_rate": 6.650732950924155e-07, + "loss": 0.0064, + "reward": 1.571221947669983, + "reward_std": 0.15910333395004272, + "rewards/pad": 0.15625, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4149720072746277, + "step": 1051 + }, + { + "completion_length": 120.046875, + "epoch": 0.3352453792224347, + "grad_norm": 11.266196250915527, + "kl": 0.1484375, + "learning_rate": 6.647546207775652e-07, + "loss": 0.0059, + "reward": 1.581168532371521, + "reward_std": 0.059135135263204575, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5811686515808105, + "step": 1052 + }, + { + "completion_length": 70.71875, + "epoch": 0.3355640535372849, + "grad_norm": 22.130834579467773, + "kl": 0.134765625, + "learning_rate": 6.64435946462715e-07, + "loss": 0.0054, + "reward": 1.530851125717163, + "reward_std": 0.07286497205495834, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5308512449264526, + "step": 1053 + }, + { + "completion_length": 174.125, + "epoch": 0.33588272785213513, + "grad_norm": 8.050396919250488, + "kl": 0.07958984375, + "learning_rate": 6.641172721478648e-07, + "loss": 0.0032, + "reward": 1.4346730709075928, + "reward_std": 0.0913335531949997, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.32529813051223755, + "rewards/pad": 0.125, + "step": 1054 + }, + { + "completion_length": 95.921875, + "epoch": 0.33620140216698535, + "grad_norm": 135.49716186523438, + "kl": 0.1591796875, + "learning_rate": 6.637985978330147e-07, + "loss": 0.0063, + "reward": 1.4473363161087036, + "reward_std": 0.09837909787893295, + "rewards/answer_reward": 0.0, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4473363161087036, + "step": 1055 + }, + { + "completion_length": 71.390625, + "epoch": 0.3365200764818356, + "grad_norm": 34.348148345947266, + "kl": 0.265625, + "learning_rate": 6.634799235181644e-07, + "loss": 0.0106, + "reward": 1.6273062229156494, + "reward_std": 0.0879427045583725, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3773062527179718, + "rewards/pad": 0.25, + "step": 1056 + }, + { + "completion_length": 44.765625, + "epoch": 0.3368387507966858, + "grad_norm": 88.11212158203125, + "kl": 0.2353515625, + "learning_rate": 6.631612492033142e-07, + "loss": 0.0094, + "reward": 1.7960689067840576, + "reward_std": 0.08121553808450699, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6710689663887024, + "rewards/pad": 0.125, + "step": 1057 + }, + { + "completion_length": 146.6875, + "epoch": 0.337157425111536, + "grad_norm": 42.112579345703125, + "kl": 0.07568359375, + "learning_rate": 6.62842574888464e-07, + "loss": 0.003, + "reward": 1.4606010913848877, + "reward_std": 0.09805289655923843, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.35122597217559814, + "step": 1058 + }, + { + "completion_length": 146.9375, + "epoch": 0.33747609942638623, + "grad_norm": 18.255258560180664, + "kl": 0.1435546875, + "learning_rate": 6.625239005736138e-07, + "loss": 0.0058, + "reward": 1.36257004737854, + "reward_std": 0.05446300283074379, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.36257001757621765, + "step": 1059 + }, + { + "completion_length": 47.015625, + "epoch": 0.33779477374123645, + "grad_norm": 32.94956588745117, + "kl": 0.16796875, + "learning_rate": 6.622052262587635e-07, + "loss": 0.0067, + "reward": 1.5549402236938477, + "reward_std": 0.10364280641078949, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5393152832984924, + "rewards/pad": 0.015625, + "step": 1060 + }, + { + "completion_length": 96.0, + "epoch": 0.3381134480560867, + "grad_norm": 19.255802154541016, + "kl": 0.16015625, + "learning_rate": 6.618865519439133e-07, + "loss": 0.0064, + "reward": 1.4055922031402588, + "reward_std": 0.0773409828543663, + "rewards/answer_reward": 0.0, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4055922329425812, + "step": 1061 + }, + { + "completion_length": 123.28125, + "epoch": 0.3384321223709369, + "grad_norm": 18.961702346801758, + "kl": 0.109375, + "learning_rate": 6.615678776290631e-07, + "loss": 0.0044, + "reward": 1.7836185693740845, + "reward_std": 0.04367794841527939, + "rewards/pad": 0.375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4086185693740845, + "step": 1062 + }, + { + "completion_length": 119.75, + "epoch": 0.3387507966857871, + "grad_norm": 20.319904327392578, + "kl": 0.0947265625, + "learning_rate": 6.612492033142129e-07, + "loss": 0.0038, + "reward": 1.5855618715286255, + "reward_std": 0.050916701555252075, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4605619013309479, + "rewards/pad": 0.125, + "step": 1063 + }, + { + "completion_length": 74.046875, + "epoch": 0.33906947100063733, + "grad_norm": 97.88804626464844, + "kl": 0.181640625, + "learning_rate": 6.609305289993626e-07, + "loss": 0.0073, + "reward": 1.7505017518997192, + "reward_std": 0.14779749512672424, + "rewards/answer_reward": 0.203125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5473767518997192, + "step": 1064 + }, + { + "completion_length": 69.875, + "epoch": 0.33938814531548755, + "grad_norm": 155.23716735839844, + "kl": 0.158203125, + "learning_rate": 6.606118546845124e-07, + "loss": 0.0063, + "reward": 1.4346399307250977, + "reward_std": 0.08623382449150085, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.43463993072509766, + "step": 1065 + }, + { + "completion_length": 94.375, + "epoch": 0.3397068196303378, + "grad_norm": 27.05531883239746, + "kl": 0.0869140625, + "learning_rate": 6.602931803696622e-07, + "loss": 0.0035, + "reward": 1.4803881645202637, + "reward_std": 0.03700976446270943, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.3553881347179413, + "step": 1066 + }, + { + "completion_length": 120.9375, + "epoch": 0.340025493945188, + "grad_norm": 12.607763290405273, + "kl": 0.177734375, + "learning_rate": 6.59974506054812e-07, + "loss": 0.0071, + "reward": 1.3900195360183716, + "reward_std": 0.03963884338736534, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3900195062160492, + "step": 1067 + }, + { + "completion_length": 70.046875, + "epoch": 0.3403441682600382, + "grad_norm": 23.681140899658203, + "kl": 0.14453125, + "learning_rate": 6.596558317399617e-07, + "loss": 0.0058, + "reward": 1.5820633172988892, + "reward_std": 0.05296536535024643, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.45706331729888916, + "rewards/pad": 0.125, + "step": 1068 + }, + { + "completion_length": 69.46875, + "epoch": 0.3406628425748885, + "grad_norm": 25.606416702270508, + "kl": 0.1953125, + "learning_rate": 6.593371574251115e-07, + "loss": 0.0078, + "reward": 1.4689515829086304, + "reward_std": 0.13900893926620483, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.46895158290863037, + "rewards/pad": 0.0, + "step": 1069 + }, + { + "completion_length": 68.84375, + "epoch": 0.3409815168897387, + "grad_norm": 51.751583099365234, + "kl": 0.12109375, + "learning_rate": 6.590184831102613e-07, + "loss": 0.0048, + "reward": 1.7762489318847656, + "reward_std": 0.10998112708330154, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5418740510940552, + "rewards/pad": 0.234375, + "step": 1070 + }, + { + "completion_length": 68.953125, + "epoch": 0.34130019120458893, + "grad_norm": 95.57342529296875, + "kl": 0.2001953125, + "learning_rate": 6.586998087954111e-07, + "loss": 0.008, + "reward": 1.5211365222930908, + "reward_std": 0.12379960715770721, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5211365222930908, + "step": 1071 + }, + { + "completion_length": 148.578125, + "epoch": 0.34161886551943915, + "grad_norm": 24.331661224365234, + "kl": 0.0830078125, + "learning_rate": 6.583811344805608e-07, + "loss": 0.0033, + "reward": 1.4323045015335083, + "reward_std": 0.16860775649547577, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.4323045015335083, + "rewards/pad": 0.015625, + "step": 1072 + }, + { + "completion_length": 97.75, + "epoch": 0.3419375398342894, + "grad_norm": 39.816951751708984, + "kl": 0.11865234375, + "learning_rate": 6.580624601657106e-07, + "loss": 0.0047, + "reward": 1.5016841888427734, + "reward_std": 0.054465897381305695, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.25168418884277344, + "step": 1073 + }, + { + "completion_length": 70.71875, + "epoch": 0.3422562141491396, + "grad_norm": 25.538747787475586, + "kl": 0.138671875, + "learning_rate": 6.577437858508605e-07, + "loss": 0.0055, + "reward": 1.491959571838379, + "reward_std": 0.0691341906785965, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.49195966124534607, + "step": 1074 + }, + { + "completion_length": 145.1875, + "epoch": 0.3425748884639898, + "grad_norm": 18.296266555786133, + "kl": 0.1259765625, + "learning_rate": 6.574251115360103e-07, + "loss": 0.005, + "reward": 1.4316325187683105, + "reward_std": 0.08940847218036652, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.43163245916366577, + "step": 1075 + }, + { + "completion_length": 70.546875, + "epoch": 0.34289356277884003, + "grad_norm": 14.101547241210938, + "kl": 0.255859375, + "learning_rate": 6.5710643722116e-07, + "loss": 0.0102, + "reward": 1.4184021949768066, + "reward_std": 0.054325029253959656, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4184022545814514, + "rewards/pad": 0.0, + "step": 1076 + }, + { + "completion_length": 43.65625, + "epoch": 0.34321223709369025, + "grad_norm": 57.003196716308594, + "kl": 0.1572265625, + "learning_rate": 6.567877629063098e-07, + "loss": 0.0063, + "reward": 1.4279489517211914, + "reward_std": 0.1659427285194397, + "rewards/pad": 0.046875, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3810740113258362, + "step": 1077 + }, + { + "completion_length": 95.5625, + "epoch": 0.3435309114085405, + "grad_norm": 51.163570404052734, + "kl": 0.1552734375, + "learning_rate": 6.564690885914596e-07, + "loss": 0.0062, + "reward": 1.669952392578125, + "reward_std": 0.07310619205236435, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5449522733688354, + "step": 1078 + }, + { + "completion_length": 44.40625, + "epoch": 0.3438495857233907, + "grad_norm": 64.22246551513672, + "kl": 0.201171875, + "learning_rate": 6.561504142766092e-07, + "loss": 0.0081, + "reward": 1.6366376876831055, + "reward_std": 0.0761965811252594, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6366376280784607, + "rewards/pad": 0.0, + "step": 1079 + }, + { + "completion_length": 44.421875, + "epoch": 0.3441682600382409, + "grad_norm": 45.51784133911133, + "kl": 0.146484375, + "learning_rate": 6.55831739961759e-07, + "loss": 0.0059, + "reward": 1.596229910850525, + "reward_std": 0.073794886469841, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4712299108505249, + "rewards/pad": 0.125, + "step": 1080 + }, + { + "completion_length": 119.171875, + "epoch": 0.34448693435309113, + "grad_norm": 26.590566635131836, + "kl": 0.111328125, + "learning_rate": 6.555130656469088e-07, + "loss": 0.0044, + "reward": 1.7527601718902588, + "reward_std": 0.11826224625110626, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.5183851718902588, + "step": 1081 + }, + { + "completion_length": 95.296875, + "epoch": 0.34480560866794135, + "grad_norm": 78.18777465820312, + "kl": 0.27734375, + "learning_rate": 6.551943913320586e-07, + "loss": 0.0111, + "reward": 1.6009079217910767, + "reward_std": 0.153416246175766, + "rewards/answer_reward": 0.046875, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5540329217910767, + "step": 1082 + }, + { + "completion_length": 70.09375, + "epoch": 0.3451242829827916, + "grad_norm": 113.71192932128906, + "kl": 0.162109375, + "learning_rate": 6.548757170172083e-07, + "loss": 0.0065, + "reward": 1.6254713535308838, + "reward_std": 0.07463856041431427, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5004714131355286, + "rewards/pad": 0.125, + "step": 1083 + }, + { + "completion_length": 46.140625, + "epoch": 0.3454429572976418, + "grad_norm": 100.79532623291016, + "kl": 0.1787109375, + "learning_rate": 6.545570427023581e-07, + "loss": 0.0072, + "reward": 1.6921272277832031, + "reward_std": 0.08855599910020828, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5671272873878479, + "rewards/pad": 0.125, + "step": 1084 + }, + { + "completion_length": 119.671875, + "epoch": 0.345761631612492, + "grad_norm": 36.70648193359375, + "kl": 0.1640625, + "learning_rate": 6.542383683875079e-07, + "loss": 0.0066, + "reward": 1.4764964580535889, + "reward_std": 0.044807374477386475, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.47649645805358887, + "step": 1085 + }, + { + "completion_length": 69.40625, + "epoch": 0.34608030592734224, + "grad_norm": 38.902225494384766, + "kl": 0.146484375, + "learning_rate": 6.539196940726577e-07, + "loss": 0.0059, + "reward": 1.5409049987792969, + "reward_std": 0.0742073804140091, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5409051179885864, + "step": 1086 + }, + { + "completion_length": 71.296875, + "epoch": 0.34639898024219246, + "grad_norm": 13.453853607177734, + "kl": 0.1572265625, + "learning_rate": 6.536010197578074e-07, + "loss": 0.0063, + "reward": 1.836655855178833, + "reward_std": 0.05411819368600845, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5866559147834778, + "rewards/pad": 0.25, + "step": 1087 + }, + { + "completion_length": 95.953125, + "epoch": 0.3467176545570427, + "grad_norm": 45.534881591796875, + "kl": 0.1064453125, + "learning_rate": 6.532823454429572e-07, + "loss": 0.0043, + "reward": 1.4753446578979492, + "reward_std": 0.052867937833070755, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.47534459829330444, + "rewards/pad": 0.0, + "step": 1088 + }, + { + "completion_length": 96.765625, + "epoch": 0.34703632887189295, + "grad_norm": 45.09857940673828, + "kl": 0.16796875, + "learning_rate": 6.52963671128107e-07, + "loss": 0.0067, + "reward": 1.6063323020935059, + "reward_std": 0.036642368882894516, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4813322424888611, + "step": 1089 + }, + { + "completion_length": 144.046875, + "epoch": 0.34735500318674317, + "grad_norm": 12.49614143371582, + "kl": 0.11572265625, + "learning_rate": 6.526449968132568e-07, + "loss": 0.0046, + "reward": 1.6987546682357788, + "reward_std": 0.07186584174633026, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4487547278404236, + "step": 1090 + }, + { + "completion_length": 145.625, + "epoch": 0.3476736775015934, + "grad_norm": 25.01397132873535, + "kl": 0.09228515625, + "learning_rate": 6.523263224984065e-07, + "loss": 0.0037, + "reward": 1.3501834869384766, + "reward_std": 0.0349920280277729, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.350183367729187, + "rewards/pad": 0.0, + "step": 1091 + }, + { + "completion_length": 120.453125, + "epoch": 0.3479923518164436, + "grad_norm": 16.605253219604492, + "kl": 0.140625, + "learning_rate": 6.520076481835563e-07, + "loss": 0.0056, + "reward": 1.478297233581543, + "reward_std": 0.13914169371128082, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.49392229318618774, + "rewards/pad": 0.0, + "step": 1092 + }, + { + "completion_length": 148.265625, + "epoch": 0.34831102613129383, + "grad_norm": 63.30807113647461, + "kl": 0.1728515625, + "learning_rate": 6.516889738687062e-07, + "loss": 0.0069, + "reward": 1.5344979763031006, + "reward_std": 0.05048166215419769, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4094979763031006, + "step": 1093 + }, + { + "completion_length": 93.75, + "epoch": 0.34862970044614405, + "grad_norm": 56.43621826171875, + "kl": 0.134765625, + "learning_rate": 6.51370299553856e-07, + "loss": 0.0054, + "reward": 1.700314998626709, + "reward_std": 0.10079171508550644, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.45031502842903137, + "step": 1094 + }, + { + "completion_length": 97.140625, + "epoch": 0.3489483747609943, + "grad_norm": 6.179287910461426, + "kl": 0.193359375, + "learning_rate": 6.510516252390057e-07, + "loss": 0.0077, + "reward": 1.6602232456207275, + "reward_std": 0.08768853545188904, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5352232456207275, + "step": 1095 + }, + { + "completion_length": 96.21875, + "epoch": 0.3492670490758445, + "grad_norm": 41.30459976196289, + "kl": 0.12158203125, + "learning_rate": 6.507329509241555e-07, + "loss": 0.0049, + "reward": 1.5766048431396484, + "reward_std": 0.07573454082012177, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4516048729419708, + "step": 1096 + }, + { + "completion_length": 95.8125, + "epoch": 0.3495857233906947, + "grad_norm": 32.63888168334961, + "kl": 0.130859375, + "learning_rate": 6.504142766093053e-07, + "loss": 0.0052, + "reward": 1.536113977432251, + "reward_std": 0.11577165871858597, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.5517390966415405, + "rewards/pad": 0.0, + "step": 1097 + }, + { + "completion_length": 143.796875, + "epoch": 0.34990439770554493, + "grad_norm": 29.69789695739746, + "kl": 0.1005859375, + "learning_rate": 6.500956022944551e-07, + "loss": 0.004, + "reward": 1.5152339935302734, + "reward_std": 0.04404383897781372, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5152339339256287, + "step": 1098 + }, + { + "completion_length": 68.65625, + "epoch": 0.35022307202039515, + "grad_norm": 23.979541778564453, + "kl": 0.1845703125, + "learning_rate": 6.497769279796048e-07, + "loss": 0.0074, + "reward": 1.4446322917938232, + "reward_std": 0.07955366373062134, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.44463226199150085, + "step": 1099 + }, + { + "completion_length": 143.3125, + "epoch": 0.3505417463352454, + "grad_norm": 23.917510986328125, + "kl": 0.1376953125, + "learning_rate": 6.494582536647546e-07, + "loss": 0.0055, + "reward": 1.5646904706954956, + "reward_std": 0.047669485211372375, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5646904110908508, + "step": 1100 + }, + { + "completion_length": 172.625, + "epoch": 0.3508604206500956, + "grad_norm": 31.693416595458984, + "kl": 0.06201171875, + "learning_rate": 6.491395793499044e-07, + "loss": 0.0025, + "reward": 1.409838080406189, + "reward_std": 0.047148752957582474, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.40983811020851135, + "step": 1101 + }, + { + "completion_length": 43.65625, + "epoch": 0.3511790949649458, + "grad_norm": 37.47905349731445, + "kl": 0.5859375, + "learning_rate": 6.488209050350542e-07, + "loss": 0.0235, + "reward": 1.8206762075424194, + "reward_std": 0.07596644014120102, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5706762671470642, + "step": 1102 + }, + { + "completion_length": 121.375, + "epoch": 0.35149776927979604, + "grad_norm": 155.16152954101562, + "kl": 0.11767578125, + "learning_rate": 6.485022307202039e-07, + "loss": 0.0047, + "reward": 1.570386528968811, + "reward_std": 0.07902577519416809, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5703864693641663, + "rewards/pad": 0.0, + "step": 1103 + }, + { + "completion_length": 146.96875, + "epoch": 0.35181644359464626, + "grad_norm": 14.959189414978027, + "kl": 0.0908203125, + "learning_rate": 6.481835564053537e-07, + "loss": 0.0036, + "reward": 1.6194053888320923, + "reward_std": 0.11047296226024628, + "rewards/pad": 0.234375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3850303590297699, + "step": 1104 + }, + { + "completion_length": 121.734375, + "epoch": 0.3521351179094965, + "grad_norm": 11.895666122436523, + "kl": 0.11376953125, + "learning_rate": 6.478648820905035e-07, + "loss": 0.0045, + "reward": 1.5688235759735107, + "reward_std": 0.04408888518810272, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5688235759735107, + "step": 1105 + }, + { + "completion_length": 169.09375, + "epoch": 0.3524537922243467, + "grad_norm": 13.154065132141113, + "kl": 0.0830078125, + "learning_rate": 6.475462077756533e-07, + "loss": 0.0033, + "reward": 1.401365041732788, + "reward_std": 0.08900520950555801, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4013649821281433, + "step": 1106 + }, + { + "completion_length": 46.984375, + "epoch": 0.3527724665391969, + "grad_norm": 49.315128326416016, + "kl": 0.2451171875, + "learning_rate": 6.47227533460803e-07, + "loss": 0.0098, + "reward": 1.7354907989501953, + "reward_std": 0.20398296415805817, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5323659181594849, + "rewards/pad": 0.203125, + "step": 1107 + }, + { + "completion_length": 98.703125, + "epoch": 0.35309114085404714, + "grad_norm": 41.61372375488281, + "kl": 0.11279296875, + "learning_rate": 6.469088591459528e-07, + "loss": 0.0045, + "reward": 1.8001890182495117, + "reward_std": 0.09147683531045914, + "rewards/pad": 0.46875, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.33143895864486694, + "step": 1108 + }, + { + "completion_length": 69.34375, + "epoch": 0.3534098151688974, + "grad_norm": 20.674589157104492, + "kl": 0.20703125, + "learning_rate": 6.465901848311026e-07, + "loss": 0.0083, + "reward": 1.4114694595336914, + "reward_std": 0.09746737778186798, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4114694595336914, + "rewards/pad": 0.0, + "step": 1109 + }, + { + "completion_length": 70.203125, + "epoch": 0.35372848948374763, + "grad_norm": 61.68789291381836, + "kl": 0.12890625, + "learning_rate": 6.462715105162523e-07, + "loss": 0.0052, + "reward": 1.512228012084961, + "reward_std": 0.07808954268693924, + "rewards/answer_reward": 0.0, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5122280716896057, + "step": 1110 + }, + { + "completion_length": 96.953125, + "epoch": 0.35404716379859785, + "grad_norm": 20.580629348754883, + "kl": 0.1220703125, + "learning_rate": 6.459528362014021e-07, + "loss": 0.0049, + "reward": 1.508711576461792, + "reward_std": 0.0672249048948288, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.508711576461792, + "rewards/pad": 0.0, + "step": 1111 + }, + { + "completion_length": 123.765625, + "epoch": 0.3543658381134481, + "grad_norm": 35.10501480102539, + "kl": 0.0654296875, + "learning_rate": 6.45634161886552e-07, + "loss": 0.0026, + "reward": 1.6096165180206299, + "reward_std": 0.04549971967935562, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3596165180206299, + "step": 1112 + }, + { + "completion_length": 70.3125, + "epoch": 0.3546845124282983, + "grad_norm": 94.42025756835938, + "kl": 0.169921875, + "learning_rate": 6.453154875717018e-07, + "loss": 0.0068, + "reward": 1.668575406074524, + "reward_std": 0.14063525199890137, + "rewards/answer_reward": 0.1875, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.48107534646987915, + "step": 1113 + }, + { + "completion_length": 93.328125, + "epoch": 0.3550031867431485, + "grad_norm": 13.738290786743164, + "kl": 0.1279296875, + "learning_rate": 6.449968132568515e-07, + "loss": 0.0051, + "reward": 1.3417246341705322, + "reward_std": 0.059449244290590286, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3417246341705322, + "rewards/pad": 0.0, + "step": 1114 + }, + { + "completion_length": 70.234375, + "epoch": 0.35532186105799873, + "grad_norm": 26.223695755004883, + "kl": 0.1279296875, + "learning_rate": 6.446781389420013e-07, + "loss": 0.0051, + "reward": 1.6727383136749268, + "reward_std": 0.05653863400220871, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.547738254070282, + "rewards/pad": 0.125, + "step": 1115 + }, + { + "completion_length": 46.046875, + "epoch": 0.35564053537284895, + "grad_norm": 252.97560119628906, + "kl": 0.1953125, + "learning_rate": 6.443594646271511e-07, + "loss": 0.0078, + "reward": 1.8712449073791504, + "reward_std": 0.10797697305679321, + "rewards/answer_reward": 0.5, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.3712449073791504, + "step": 1116 + }, + { + "completion_length": 68.796875, + "epoch": 0.3559592096876992, + "grad_norm": 14.689714431762695, + "kl": 0.12890625, + "learning_rate": 6.440407903123009e-07, + "loss": 0.0051, + "reward": 1.6983696222305298, + "reward_std": 0.05995948985219002, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.44836968183517456, + "step": 1117 + }, + { + "completion_length": 69.078125, + "epoch": 0.3562778840025494, + "grad_norm": 27.807676315307617, + "kl": 0.197265625, + "learning_rate": 6.437221159974505e-07, + "loss": 0.0079, + "reward": 1.6848386526107788, + "reward_std": 0.07083934545516968, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5598386526107788, + "step": 1118 + }, + { + "completion_length": 150.578125, + "epoch": 0.3565965583173996, + "grad_norm": 17.879209518432617, + "kl": 0.09765625, + "learning_rate": 6.434034416826003e-07, + "loss": 0.0039, + "reward": 1.50717031955719, + "reward_std": 0.07021446526050568, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.38217031955718994, + "step": 1119 + }, + { + "completion_length": 95.578125, + "epoch": 0.35691523263224983, + "grad_norm": 39.17152404785156, + "kl": 0.130859375, + "learning_rate": 6.430847673677501e-07, + "loss": 0.0052, + "reward": 1.4707567691802979, + "reward_std": 0.02951684780418873, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.47075676918029785, + "rewards/pad": 0.0, + "step": 1120 + }, + { + "completion_length": 71.09375, + "epoch": 0.35723390694710005, + "grad_norm": 20.048585891723633, + "kl": 0.2021484375, + "learning_rate": 6.427660930528999e-07, + "loss": 0.0081, + "reward": 1.9080220460891724, + "reward_std": 0.13876330852508545, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5486471056938171, + "rewards/pad": 0.359375, + "step": 1121 + }, + { + "completion_length": 71.546875, + "epoch": 0.3575525812619503, + "grad_norm": 36.162559509277344, + "kl": 0.1484375, + "learning_rate": 6.424474187380496e-07, + "loss": 0.0059, + "reward": 1.79642915725708, + "reward_std": 0.07941967248916626, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5464290976524353, + "rewards/pad": 0.25, + "step": 1122 + }, + { + "completion_length": 94.0625, + "epoch": 0.3578712555768005, + "grad_norm": 42.912445068359375, + "kl": 0.1181640625, + "learning_rate": 6.421287444231994e-07, + "loss": 0.0047, + "reward": 1.6768763065338135, + "reward_std": 0.06940555572509766, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6768762469291687, + "rewards/pad": 0.0, + "step": 1123 + }, + { + "completion_length": 94.75, + "epoch": 0.3581899298916507, + "grad_norm": 39.69688415527344, + "kl": 0.1240234375, + "learning_rate": 6.418100701083492e-07, + "loss": 0.005, + "reward": 1.3529894351959229, + "reward_std": 0.049606405198574066, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3529894948005676, + "rewards/pad": 0.0, + "step": 1124 + }, + { + "completion_length": 147.765625, + "epoch": 0.35850860420650094, + "grad_norm": 25.249536514282227, + "kl": 0.08837890625, + "learning_rate": 6.41491395793499e-07, + "loss": 0.0035, + "reward": 1.538456678390503, + "reward_std": 0.0675881952047348, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4134565591812134, + "rewards/pad": 0.125, + "step": 1125 + }, + { + "completion_length": 72.0625, + "epoch": 0.35882727852135116, + "grad_norm": 115.1136703491211, + "kl": 0.12451171875, + "learning_rate": 6.411727214786487e-07, + "loss": 0.005, + "reward": 1.5841870307922363, + "reward_std": 0.10195166617631912, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.45918700098991394, + "rewards/pad": 0.125, + "step": 1126 + }, + { + "completion_length": 72.90625, + "epoch": 0.3591459528362014, + "grad_norm": 18.511503219604492, + "kl": 0.185546875, + "learning_rate": 6.408540471637985e-07, + "loss": 0.0074, + "reward": 1.6491913795471191, + "reward_std": 0.09406599402427673, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.39919140934944153, + "step": 1127 + }, + { + "completion_length": 173.046875, + "epoch": 0.35946462715105165, + "grad_norm": 20.802494049072266, + "kl": 0.107421875, + "learning_rate": 6.405353728489483e-07, + "loss": 0.0043, + "reward": 1.3294429779052734, + "reward_std": 0.048001088201999664, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.32944291830062866, + "step": 1128 + }, + { + "completion_length": 71.421875, + "epoch": 0.3597833014659019, + "grad_norm": 50.930999755859375, + "kl": 0.423828125, + "learning_rate": 6.402166985340981e-07, + "loss": 0.0169, + "reward": 1.7195351123809814, + "reward_std": 0.10376234352588654, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5945351123809814, + "step": 1129 + }, + { + "completion_length": 96.796875, + "epoch": 0.3601019757807521, + "grad_norm": 17.261219024658203, + "kl": 0.216796875, + "learning_rate": 6.398980242192478e-07, + "loss": 0.0086, + "reward": 1.6543177366256714, + "reward_std": 0.14827221632003784, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5293177366256714, + "rewards/pad": 0.125, + "step": 1130 + }, + { + "completion_length": 122.40625, + "epoch": 0.3604206500956023, + "grad_norm": 7.444701671600342, + "kl": 0.08544921875, + "learning_rate": 6.395793499043977e-07, + "loss": 0.0034, + "reward": 1.3980233669281006, + "reward_std": 0.04345095157623291, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.273023396730423, + "rewards/pad": 0.125, + "step": 1131 + }, + { + "completion_length": 95.921875, + "epoch": 0.36073932441045253, + "grad_norm": 17.02141571044922, + "kl": 0.109375, + "learning_rate": 6.392606755895475e-07, + "loss": 0.0044, + "reward": 1.536940574645996, + "reward_std": 0.03979106992483139, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5369405150413513, + "rewards/pad": 0.0, + "step": 1132 + }, + { + "completion_length": 100.96875, + "epoch": 0.36105799872530275, + "grad_norm": 26.141685485839844, + "kl": 0.1875, + "learning_rate": 6.389420012746973e-07, + "loss": 0.0075, + "reward": 1.7413113117218018, + "reward_std": 0.07656162232160568, + "rewards/answer_reward": 0.375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.36631134152412415, + "step": 1133 + }, + { + "completion_length": 46.203125, + "epoch": 0.361376673040153, + "grad_norm": 37.33512878417969, + "kl": 0.20703125, + "learning_rate": 6.38623326959847e-07, + "loss": 0.0083, + "reward": 1.7170262336730957, + "reward_std": 0.11353346705436707, + "rewards/answer_reward": 0.21875, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4982762336730957, + "step": 1134 + }, + { + "completion_length": 121.359375, + "epoch": 0.3616953473550032, + "grad_norm": 16.106428146362305, + "kl": 0.103515625, + "learning_rate": 6.383046526449968e-07, + "loss": 0.0041, + "reward": 1.5457892417907715, + "reward_std": 0.1631377637386322, + "rewards/format_reward_tg": 0.96875, + "rewards/iou_timestamp_reward": 0.5770392417907715, + "rewards/pad": 0.0, + "step": 1135 + }, + { + "completion_length": 19.734375, + "epoch": 0.3620140216698534, + "grad_norm": 43.21052169799805, + "kl": 0.138671875, + "learning_rate": 6.379859783301466e-07, + "loss": 0.0056, + "reward": 1.8942912817001343, + "reward_std": 0.07131467759609222, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6442912220954895, + "rewards/pad": 0.25, + "step": 1136 + }, + { + "completion_length": 122.40625, + "epoch": 0.36233269598470363, + "grad_norm": 27.354520797729492, + "kl": 0.1123046875, + "learning_rate": 6.376673040152964e-07, + "loss": 0.0045, + "reward": 1.372056245803833, + "reward_std": 0.0385003387928009, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.372056245803833, + "rewards/pad": 0.0, + "step": 1137 + }, + { + "completion_length": 46.96875, + "epoch": 0.36265137029955385, + "grad_norm": 91.3863296508789, + "kl": 0.1689453125, + "learning_rate": 6.373486297004461e-07, + "loss": 0.0067, + "reward": 1.6147258281707764, + "reward_std": 0.1302720457315445, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.47410082817077637, + "rewards/pad": 0.140625, + "step": 1138 + }, + { + "completion_length": 144.875, + "epoch": 0.3629700446144041, + "grad_norm": 15.614913940429688, + "kl": 0.099609375, + "learning_rate": 6.370299553855959e-07, + "loss": 0.004, + "reward": 1.4381680488586426, + "reward_std": 0.051587171852588654, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4381680488586426, + "step": 1139 + }, + { + "completion_length": 74.890625, + "epoch": 0.3632887189292543, + "grad_norm": 26.624767303466797, + "kl": 0.14453125, + "learning_rate": 6.367112810707457e-07, + "loss": 0.0058, + "reward": 1.7303261756896973, + "reward_std": 0.05172932520508766, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.48032620549201965, + "rewards/pad": 0.25, + "step": 1140 + }, + { + "completion_length": 123.015625, + "epoch": 0.3636073932441045, + "grad_norm": 52.83194351196289, + "kl": 0.12158203125, + "learning_rate": 6.363926067558954e-07, + "loss": 0.0049, + "reward": 1.5532702207565308, + "reward_std": 0.07574497163295746, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.553270161151886, + "step": 1141 + }, + { + "completion_length": 43.84375, + "epoch": 0.36392606755895474, + "grad_norm": 18.857934951782227, + "kl": 0.23828125, + "learning_rate": 6.360739324410452e-07, + "loss": 0.0095, + "reward": 1.6999969482421875, + "reward_std": 0.1190653145313263, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5749968886375427, + "rewards/pad": 0.125, + "step": 1142 + }, + { + "completion_length": 72.453125, + "epoch": 0.36424474187380496, + "grad_norm": 49.3443603515625, + "kl": 0.2236328125, + "learning_rate": 6.35755258126195e-07, + "loss": 0.0089, + "reward": 1.822791576385498, + "reward_std": 0.11447308957576752, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5727915167808533, + "rewards/pad": 0.25, + "step": 1143 + }, + { + "completion_length": 69.9375, + "epoch": 0.3645634161886552, + "grad_norm": 23.304380416870117, + "kl": 0.1923828125, + "learning_rate": 6.354365838113448e-07, + "loss": 0.0077, + "reward": 1.5509254932403564, + "reward_std": 0.11732175946235657, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.550925612449646, + "rewards/pad": 0.0, + "step": 1144 + }, + { + "completion_length": 45.34375, + "epoch": 0.3648820905035054, + "grad_norm": 21.73938751220703, + "kl": 0.2109375, + "learning_rate": 6.351179094964945e-07, + "loss": 0.0084, + "reward": 1.6023046970367432, + "reward_std": 0.08304779976606369, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6023047566413879, + "rewards/pad": 0.0, + "step": 1145 + }, + { + "completion_length": 71.578125, + "epoch": 0.3652007648183556, + "grad_norm": 45.14289474487305, + "kl": 0.1376953125, + "learning_rate": 6.347992351816443e-07, + "loss": 0.0055, + "reward": 1.6271437406539917, + "reward_std": 0.11034458130598068, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5177686810493469, + "rewards/pad": 0.109375, + "step": 1146 + }, + { + "completion_length": 122.375, + "epoch": 0.36551943913320584, + "grad_norm": 45.17229461669922, + "kl": 0.0947265625, + "learning_rate": 6.344805608667941e-07, + "loss": 0.0038, + "reward": 1.6036796569824219, + "reward_std": 0.06949443370103836, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.47867974638938904, + "step": 1147 + }, + { + "completion_length": 95.96875, + "epoch": 0.3658381134480561, + "grad_norm": 23.348485946655273, + "kl": 0.1748046875, + "learning_rate": 6.34161886551944e-07, + "loss": 0.007, + "reward": 1.8184211254119873, + "reward_std": 0.07187941670417786, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.6934211254119873, + "step": 1148 + }, + { + "completion_length": 98.015625, + "epoch": 0.36615678776290633, + "grad_norm": 51.460044860839844, + "kl": 0.1103515625, + "learning_rate": 6.338432122370936e-07, + "loss": 0.0044, + "reward": 1.6390737295150757, + "reward_std": 0.11423532664775848, + "rewards/pad": 0.140625, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.49844881892204285, + "step": 1149 + }, + { + "completion_length": 123.265625, + "epoch": 0.36647546207775655, + "grad_norm": 35.537864685058594, + "kl": 0.1181640625, + "learning_rate": 6.335245379222435e-07, + "loss": 0.0047, + "reward": 1.4418638944625854, + "reward_std": 0.1120540201663971, + "rewards/answer_reward": 0.078125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.36373889446258545, + "step": 1150 + }, + { + "completion_length": 45.078125, + "epoch": 0.3667941363926068, + "grad_norm": 85.31786346435547, + "kl": 0.140625, + "learning_rate": 6.332058636073933e-07, + "loss": 0.0056, + "reward": 1.5280697345733643, + "reward_std": 0.06672141700983047, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5280698537826538, + "rewards/pad": 0.0, + "step": 1151 + }, + { + "completion_length": 124.5, + "epoch": 0.367112810707457, + "grad_norm": 18.085052490234375, + "kl": 0.08642578125, + "learning_rate": 6.328871892925431e-07, + "loss": 0.0035, + "reward": 1.5425379276275635, + "reward_std": 0.047018010169267654, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4175378978252411, + "step": 1152 + }, + { + "completion_length": 122.03125, + "epoch": 0.3674314850223072, + "grad_norm": 41.6903076171875, + "kl": 0.2177734375, + "learning_rate": 6.325685149776928e-07, + "loss": 0.0087, + "reward": 1.4459846019744873, + "reward_std": 0.09456399083137512, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4459846615791321, + "rewards/pad": 0.0, + "step": 1153 + }, + { + "completion_length": 71.5, + "epoch": 0.36775015933715743, + "grad_norm": 16.231029510498047, + "kl": 0.1591796875, + "learning_rate": 6.322498406628426e-07, + "loss": 0.0064, + "reward": 1.67655611038208, + "reward_std": 0.11449247598648071, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5515559911727905, + "step": 1154 + }, + { + "completion_length": 97.703125, + "epoch": 0.36806883365200765, + "grad_norm": 14.319849967956543, + "kl": 0.095703125, + "learning_rate": 6.319311663479924e-07, + "loss": 0.0038, + "reward": 1.6368542909622192, + "reward_std": 0.08659229427576065, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6368542909622192, + "rewards/pad": 0.0, + "step": 1155 + }, + { + "completion_length": 98.09375, + "epoch": 0.3683875079668579, + "grad_norm": 27.879743576049805, + "kl": 0.08203125, + "learning_rate": 6.316124920331422e-07, + "loss": 0.0033, + "reward": 1.854182481765747, + "reward_std": 0.13125072419643402, + "rewards/pad": 0.5, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.3698073625564575, + "step": 1156 + }, + { + "completion_length": 98.9375, + "epoch": 0.3687061822817081, + "grad_norm": 71.60199737548828, + "kl": 0.1923828125, + "learning_rate": 6.312938177182919e-07, + "loss": 0.0077, + "reward": 1.548559546470642, + "reward_std": 0.05056636035442352, + "rewards/answer_reward": 0.0, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5485595464706421, + "step": 1157 + }, + { + "completion_length": 52.703125, + "epoch": 0.3690248565965583, + "grad_norm": 134.50344848632812, + "kl": 0.142578125, + "learning_rate": 6.309751434034416e-07, + "loss": 0.0057, + "reward": 1.6984164714813232, + "reward_std": 0.13021734356880188, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.5890415906906128, + "rewards/pad": 0.125, + "step": 1158 + }, + { + "completion_length": 147.734375, + "epoch": 0.36934353091140854, + "grad_norm": 50.460662841796875, + "kl": 0.10205078125, + "learning_rate": 6.306564690885914e-07, + "loss": 0.0041, + "reward": 1.4703478813171387, + "reward_std": 0.04957844689488411, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.47034788131713867, + "step": 1159 + }, + { + "completion_length": 71.421875, + "epoch": 0.36966220522625876, + "grad_norm": 28.34197425842285, + "kl": 0.1484375, + "learning_rate": 6.303377947737412e-07, + "loss": 0.0059, + "reward": 1.50392746925354, + "reward_std": 0.05638600140810013, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.3789275586605072, + "step": 1160 + }, + { + "completion_length": 70.875, + "epoch": 0.369980879541109, + "grad_norm": 104.68994903564453, + "kl": 0.212890625, + "learning_rate": 6.300191204588909e-07, + "loss": 0.0085, + "reward": 1.6136631965637207, + "reward_std": 0.08198197185993195, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4886631369590759, + "rewards/pad": 0.125, + "step": 1161 + }, + { + "completion_length": 121.9375, + "epoch": 0.3702995538559592, + "grad_norm": 58.8842658996582, + "kl": 0.1083984375, + "learning_rate": 6.297004461440407e-07, + "loss": 0.0043, + "reward": 1.454060673713684, + "reward_std": 0.08632092922925949, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4540606439113617, + "rewards/pad": 0.0, + "step": 1162 + }, + { + "completion_length": 94.828125, + "epoch": 0.3706182281708094, + "grad_norm": 27.434877395629883, + "kl": 0.37109375, + "learning_rate": 6.293817718291905e-07, + "loss": 0.0149, + "reward": 1.4339649677276611, + "reward_std": 0.10790198296308517, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4339650273323059, + "rewards/pad": 0.0, + "step": 1163 + }, + { + "completion_length": 149.75, + "epoch": 0.37093690248565964, + "grad_norm": 19.707164764404297, + "kl": 0.08740234375, + "learning_rate": 6.290630975143403e-07, + "loss": 0.0035, + "reward": 1.531341552734375, + "reward_std": 0.04004561901092529, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.531341552734375, + "step": 1164 + }, + { + "completion_length": 72.828125, + "epoch": 0.37125557680050986, + "grad_norm": 88.5399169921875, + "kl": 0.10400390625, + "learning_rate": 6.2874442319949e-07, + "loss": 0.0042, + "reward": 1.9113043546676636, + "reward_std": 0.23165732622146606, + "rewards/answer_reward": 0.359375, + "rewards/format_reward_gqa": 0.96875, + "rewards/iou_glue_reward": 0.5831793546676636, + "step": 1165 + }, + { + "completion_length": 149.5, + "epoch": 0.3715742511153601, + "grad_norm": 16.100812911987305, + "kl": 0.181640625, + "learning_rate": 6.284257488846398e-07, + "loss": 0.0073, + "reward": 1.3362720012664795, + "reward_std": 0.04164520651102066, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.33627191185951233, + "rewards/pad": 0.0, + "step": 1166 + }, + { + "completion_length": 95.109375, + "epoch": 0.37189292543021035, + "grad_norm": 20.495891571044922, + "kl": 0.1298828125, + "learning_rate": 6.281070745697896e-07, + "loss": 0.0052, + "reward": 1.5656877756118774, + "reward_std": 0.047522690147161484, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5656878352165222, + "rewards/pad": 0.0, + "step": 1167 + }, + { + "completion_length": 69.09375, + "epoch": 0.3722115997450606, + "grad_norm": 33.39354705810547, + "kl": 0.1572265625, + "learning_rate": 6.277884002549393e-07, + "loss": 0.0063, + "reward": 1.602851390838623, + "reward_std": 0.13379782438278198, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.618476390838623, + "step": 1168 + }, + { + "completion_length": 70.90625, + "epoch": 0.3725302740599108, + "grad_norm": 88.5570068359375, + "kl": 0.1943359375, + "learning_rate": 6.274697259400892e-07, + "loss": 0.0078, + "reward": 1.6135051250457764, + "reward_std": 0.1037006676197052, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.48850518465042114, + "step": 1169 + }, + { + "completion_length": 98.890625, + "epoch": 0.372848948374761, + "grad_norm": 27.904115676879883, + "kl": 0.166015625, + "learning_rate": 6.27151051625239e-07, + "loss": 0.0066, + "reward": 1.5022716522216797, + "reward_std": 0.06810413300991058, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.37727171182632446, + "rewards/pad": 0.125, + "step": 1170 + }, + { + "completion_length": 45.5625, + "epoch": 0.37316762268961123, + "grad_norm": 35.16514587402344, + "kl": 0.2216796875, + "learning_rate": 6.268323773103888e-07, + "loss": 0.0089, + "reward": 1.522203803062439, + "reward_std": 0.1130586639046669, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5065787434577942, + "rewards/pad": 0.015625, + "step": 1171 + }, + { + "completion_length": 122.375, + "epoch": 0.37348629700446145, + "grad_norm": 152.366455078125, + "kl": 0.10693359375, + "learning_rate": 6.265137029955385e-07, + "loss": 0.0043, + "reward": 1.3803231716156006, + "reward_std": 0.08589351922273636, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.2709481418132782, + "step": 1172 + }, + { + "completion_length": 72.640625, + "epoch": 0.3738049713193117, + "grad_norm": 19.461612701416016, + "kl": 0.205078125, + "learning_rate": 6.261950286806883e-07, + "loss": 0.0082, + "reward": 1.5194697380065918, + "reward_std": 0.09573933482170105, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.519469678401947, + "rewards/pad": 0.0, + "step": 1173 + }, + { + "completion_length": 121.671875, + "epoch": 0.3741236456341619, + "grad_norm": 21.8033390045166, + "kl": 0.12109375, + "learning_rate": 6.258763543658381e-07, + "loss": 0.0048, + "reward": 1.5664465427398682, + "reward_std": 0.10758214443922043, + "rewards/pad": 0.03125, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.5508216619491577, + "step": 1174 + }, + { + "completion_length": 149.265625, + "epoch": 0.3744423199490121, + "grad_norm": 76.57758331298828, + "kl": 0.11572265625, + "learning_rate": 6.255576800509879e-07, + "loss": 0.0046, + "reward": 1.3753361701965332, + "reward_std": 0.0831504836678505, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3753361701965332, + "step": 1175 + }, + { + "completion_length": 70.6875, + "epoch": 0.37476099426386233, + "grad_norm": 37.332637786865234, + "kl": 0.203125, + "learning_rate": 6.252390057361376e-07, + "loss": 0.0081, + "reward": 1.3779017925262451, + "reward_std": 0.11319969594478607, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3779017925262451, + "step": 1176 + }, + { + "completion_length": 71.0625, + "epoch": 0.37507966857871256, + "grad_norm": 33.94636535644531, + "kl": 0.1513671875, + "learning_rate": 6.249203314212874e-07, + "loss": 0.0061, + "reward": 1.6451020240783691, + "reward_std": 0.14212313294410706, + "rewards/pad": 0.046875, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5982270240783691, + "step": 1177 + }, + { + "completion_length": 44.984375, + "epoch": 0.3753983428935628, + "grad_norm": 32.17403030395508, + "kl": 0.1904296875, + "learning_rate": 6.246016571064372e-07, + "loss": 0.0076, + "reward": 1.541552186012268, + "reward_std": 0.0988098680973053, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.41655224561691284, + "rewards/pad": 0.125, + "step": 1178 + }, + { + "completion_length": 121.046875, + "epoch": 0.375717017208413, + "grad_norm": 14.725617408752441, + "kl": 0.1328125, + "learning_rate": 6.24282982791587e-07, + "loss": 0.0053, + "reward": 1.4356043338775635, + "reward_std": 0.0906897485256195, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.451229453086853, + "step": 1179 + }, + { + "completion_length": 151.296875, + "epoch": 0.3760356915232632, + "grad_norm": 13.176466941833496, + "kl": 0.09765625, + "learning_rate": 6.239643084767367e-07, + "loss": 0.0039, + "reward": 1.492295265197754, + "reward_std": 0.09461408853530884, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.5079202651977539, + "step": 1180 + }, + { + "completion_length": 97.21875, + "epoch": 0.37635436583811344, + "grad_norm": 38.49825668334961, + "kl": 0.1787109375, + "learning_rate": 6.236456341618865e-07, + "loss": 0.0072, + "reward": 1.4638856649398804, + "reward_std": 0.04449250549077988, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.46388566493988037, + "step": 1181 + }, + { + "completion_length": 148.8125, + "epoch": 0.37667304015296366, + "grad_norm": 21.729101181030273, + "kl": 0.07763671875, + "learning_rate": 6.233269598470363e-07, + "loss": 0.0031, + "reward": 1.580154299736023, + "reward_std": 0.029881590977311134, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4551542401313782, + "step": 1182 + }, + { + "completion_length": 150.09375, + "epoch": 0.3769917144678139, + "grad_norm": 25.908597946166992, + "kl": 0.07421875, + "learning_rate": 6.230082855321861e-07, + "loss": 0.003, + "reward": 1.4306869506835938, + "reward_std": 0.032189082354307175, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4306868314743042, + "step": 1183 + }, + { + "completion_length": 99.0, + "epoch": 0.3773103887826641, + "grad_norm": 25.995725631713867, + "kl": 0.09716796875, + "learning_rate": 6.226896112173358e-07, + "loss": 0.0039, + "reward": 1.934666395187378, + "reward_std": 0.05638587474822998, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.6846664547920227, + "step": 1184 + }, + { + "completion_length": 96.15625, + "epoch": 0.3776290630975143, + "grad_norm": 36.28327560424805, + "kl": 0.173828125, + "learning_rate": 6.223709369024856e-07, + "loss": 0.0069, + "reward": 1.549536943435669, + "reward_std": 0.14491955935955048, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5495370626449585, + "rewards/pad": 0.0, + "step": 1185 + }, + { + "completion_length": 125.859375, + "epoch": 0.37794773741236454, + "grad_norm": 57.64690399169922, + "kl": 0.13671875, + "learning_rate": 6.220522625876354e-07, + "loss": 0.0055, + "reward": 1.4673360586166382, + "reward_std": 0.062283311039209366, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.3423360586166382, + "step": 1186 + }, + { + "completion_length": 45.984375, + "epoch": 0.3782664117272148, + "grad_norm": 64.22225189208984, + "kl": 0.1513671875, + "learning_rate": 6.217335882727853e-07, + "loss": 0.006, + "reward": 1.6241674423217773, + "reward_std": 0.09040925651788712, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6241674423217773, + "rewards/pad": 0.0, + "step": 1187 + }, + { + "completion_length": 71.9375, + "epoch": 0.37858508604206503, + "grad_norm": 114.3721923828125, + "kl": 0.1259765625, + "learning_rate": 6.21414913957935e-07, + "loss": 0.005, + "reward": 1.732898473739624, + "reward_std": 0.12032415717840195, + "rewards/answer_reward": 0.21875, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.514148473739624, + "step": 1188 + }, + { + "completion_length": 97.3125, + "epoch": 0.37890376035691525, + "grad_norm": 35.77212142944336, + "kl": 0.1298828125, + "learning_rate": 6.210962396430848e-07, + "loss": 0.0052, + "reward": 1.8220728635787964, + "reward_std": 0.1216290220618248, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5876979231834412, + "rewards/pad": 0.234375, + "step": 1189 + }, + { + "completion_length": 120.6875, + "epoch": 0.3792224346717655, + "grad_norm": 23.409149169921875, + "kl": 0.11474609375, + "learning_rate": 6.207775653282346e-07, + "loss": 0.0046, + "reward": 1.434619426727295, + "reward_std": 0.04008358716964722, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.43461936712265015, + "step": 1190 + }, + { + "completion_length": 124.96875, + "epoch": 0.3795411089866157, + "grad_norm": 38.4090576171875, + "kl": 0.1337890625, + "learning_rate": 6.204588910133844e-07, + "loss": 0.0053, + "reward": 1.5721570253372192, + "reward_std": 0.045577432960271835, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.44715702533721924, + "rewards/pad": 0.125, + "step": 1191 + }, + { + "completion_length": 17.8125, + "epoch": 0.3798597833014659, + "grad_norm": 47.2709846496582, + "kl": 0.1533203125, + "learning_rate": 6.201402166985341e-07, + "loss": 0.0061, + "reward": 1.6977386474609375, + "reward_std": 0.048093028366565704, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.697738766670227, + "rewards/pad": 0.0, + "step": 1192 + }, + { + "completion_length": 150.890625, + "epoch": 0.38017845761631613, + "grad_norm": 17.609628677368164, + "kl": 0.08056640625, + "learning_rate": 6.198215423836839e-07, + "loss": 0.0032, + "reward": 1.5538386106491089, + "reward_std": 0.10176517069339752, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.5694636106491089, + "step": 1193 + }, + { + "completion_length": 98.546875, + "epoch": 0.38049713193116635, + "grad_norm": 36.28522872924805, + "kl": 0.466796875, + "learning_rate": 6.195028680688337e-07, + "loss": 0.0186, + "reward": 1.6114730834960938, + "reward_std": 0.07779565453529358, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.48647308349609375, + "step": 1194 + }, + { + "completion_length": 72.40625, + "epoch": 0.3808158062460166, + "grad_norm": 50.89261245727539, + "kl": 0.2353515625, + "learning_rate": 6.191841937539835e-07, + "loss": 0.0094, + "reward": 1.6488547325134277, + "reward_std": 0.14085832238197327, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5082297921180725, + "rewards/pad": 0.140625, + "step": 1195 + }, + { + "completion_length": 71.03125, + "epoch": 0.3811344805608668, + "grad_norm": 41.0035514831543, + "kl": 0.150390625, + "learning_rate": 6.188655194391332e-07, + "loss": 0.006, + "reward": 1.6350042819976807, + "reward_std": 0.05474071204662323, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6350042819976807, + "rewards/pad": 0.0, + "step": 1196 + }, + { + "completion_length": 45.1875, + "epoch": 0.381453154875717, + "grad_norm": 38.76614761352539, + "kl": 0.298828125, + "learning_rate": 6.185468451242829e-07, + "loss": 0.012, + "reward": 1.6334543228149414, + "reward_std": 0.07611322402954102, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6334543228149414, + "rewards/pad": 0.0, + "step": 1197 + }, + { + "completion_length": 98.0625, + "epoch": 0.38177182919056724, + "grad_norm": 15.111804962158203, + "kl": 0.1328125, + "learning_rate": 6.182281708094327e-07, + "loss": 0.0053, + "reward": 1.5155563354492188, + "reward_std": 0.040137313306331635, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5155563354492188, + "step": 1198 + }, + { + "completion_length": 152.421875, + "epoch": 0.38209050350541746, + "grad_norm": 27.031673431396484, + "kl": 0.169921875, + "learning_rate": 6.179094964945824e-07, + "loss": 0.0068, + "reward": 1.5351272821426392, + "reward_std": 0.10124079883098602, + "rewards/answer_reward": 0.140625, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.3945022225379944, + "step": 1199 + }, + { + "completion_length": 98.328125, + "epoch": 0.3824091778202677, + "grad_norm": 64.51046752929688, + "kl": 0.1982421875, + "learning_rate": 6.175908221797322e-07, + "loss": 0.0079, + "reward": 1.6580084562301636, + "reward_std": 0.11241358518600464, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4861334562301636, + "rewards/pad": 0.171875, + "step": 1200 + }, + { + "completion_length": 150.921875, + "epoch": 0.3827278521351179, + "grad_norm": 30.153474807739258, + "kl": 0.09033203125, + "learning_rate": 6.17272147864882e-07, + "loss": 0.0036, + "reward": 1.6002533435821533, + "reward_std": 0.04925467073917389, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.47525322437286377, + "step": 1201 + }, + { + "completion_length": 123.421875, + "epoch": 0.3830465264499681, + "grad_norm": 12.115293502807617, + "kl": 0.115234375, + "learning_rate": 6.169534735500318e-07, + "loss": 0.0046, + "reward": 1.4110045433044434, + "reward_std": 0.035006508231163025, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.28600451350212097, + "step": 1202 + }, + { + "completion_length": 147.078125, + "epoch": 0.38336520076481834, + "grad_norm": 10.549017906188965, + "kl": 0.11474609375, + "learning_rate": 6.166347992351815e-07, + "loss": 0.0046, + "reward": 1.5266410112380981, + "reward_std": 0.07672546058893204, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5266410112380981, + "step": 1203 + }, + { + "completion_length": 70.015625, + "epoch": 0.38368387507966856, + "grad_norm": 71.49323272705078, + "kl": 0.1259765625, + "learning_rate": 6.163161249203313e-07, + "loss": 0.0051, + "reward": 1.581700086593628, + "reward_std": 0.11010384559631348, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5192000865936279, + "rewards/pad": 0.0625, + "step": 1204 + }, + { + "completion_length": 150.140625, + "epoch": 0.3840025493945188, + "grad_norm": 19.70912742614746, + "kl": 0.08984375, + "learning_rate": 6.159974506054811e-07, + "loss": 0.0036, + "reward": 1.5954060554504395, + "reward_std": 0.06902284920215607, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4704059660434723, + "step": 1205 + }, + { + "completion_length": 122.296875, + "epoch": 0.384321223709369, + "grad_norm": 16.765697479248047, + "kl": 0.11962890625, + "learning_rate": 6.15678776290631e-07, + "loss": 0.0048, + "reward": 1.4820270538330078, + "reward_std": 0.07253751158714294, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.372652143239975, + "step": 1206 + }, + { + "completion_length": 20.453125, + "epoch": 0.3846398980242193, + "grad_norm": 39.035213470458984, + "kl": 0.201171875, + "learning_rate": 6.153601019757807e-07, + "loss": 0.008, + "reward": 1.7460757493972778, + "reward_std": 0.05062546581029892, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.49607568979263306, + "step": 1207 + }, + { + "completion_length": 72.0, + "epoch": 0.3849585723390695, + "grad_norm": 78.69184112548828, + "kl": 0.146484375, + "learning_rate": 6.150414276609305e-07, + "loss": 0.0058, + "reward": 1.5769894123077393, + "reward_std": 0.18609334528446198, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4988645017147064, + "rewards/pad": 0.078125, + "step": 1208 + }, + { + "completion_length": 123.75, + "epoch": 0.3852772466539197, + "grad_norm": 49.43526840209961, + "kl": 0.1669921875, + "learning_rate": 6.147227533460803e-07, + "loss": 0.0067, + "reward": 1.592034101486206, + "reward_std": 0.051229365170001984, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.46703410148620605, + "rewards/pad": 0.125, + "step": 1209 + }, + { + "completion_length": 122.46875, + "epoch": 0.38559592096876993, + "grad_norm": 13.823474884033203, + "kl": 0.1748046875, + "learning_rate": 6.144040790312301e-07, + "loss": 0.007, + "reward": 1.4730191230773926, + "reward_std": 0.11505133658647537, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.4886440932750702, + "step": 1210 + }, + { + "completion_length": 70.1875, + "epoch": 0.38591459528362015, + "grad_norm": 30.62668228149414, + "kl": 0.1640625, + "learning_rate": 6.140854047163798e-07, + "loss": 0.0066, + "reward": 1.5743595361709595, + "reward_std": 0.06021110713481903, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5743595361709595, + "step": 1211 + }, + { + "completion_length": 123.390625, + "epoch": 0.3862332695984704, + "grad_norm": 15.12385082244873, + "kl": 0.12451171875, + "learning_rate": 6.137667304015296e-07, + "loss": 0.005, + "reward": 1.4223942756652832, + "reward_std": 0.04839033633470535, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4223942756652832, + "rewards/pad": 0.0, + "step": 1212 + }, + { + "completion_length": 66.296875, + "epoch": 0.3865519439133206, + "grad_norm": 37.55725860595703, + "kl": 0.357421875, + "learning_rate": 6.134480560866794e-07, + "loss": 0.0142, + "reward": 1.6173839569091797, + "reward_std": 0.12178170680999756, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5548838973045349, + "rewards/pad": 0.0625, + "step": 1213 + }, + { + "completion_length": 69.59375, + "epoch": 0.3868706182281708, + "grad_norm": 67.85687255859375, + "kl": 0.228515625, + "learning_rate": 6.131293817718292e-07, + "loss": 0.0091, + "reward": 1.5872609615325928, + "reward_std": 0.11316142231225967, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.47788599133491516, + "rewards/pad": 0.125, + "step": 1214 + }, + { + "completion_length": 123.65625, + "epoch": 0.38718929254302104, + "grad_norm": 22.573434829711914, + "kl": 0.171875, + "learning_rate": 6.128107074569789e-07, + "loss": 0.0069, + "reward": 1.4954043626785278, + "reward_std": 0.04548972100019455, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.49540436267852783, + "step": 1215 + }, + { + "completion_length": 173.71875, + "epoch": 0.38750796685787126, + "grad_norm": 8.874141693115234, + "kl": 0.059814453125, + "learning_rate": 6.124920331421287e-07, + "loss": 0.0024, + "reward": 1.5322620868682861, + "reward_std": 0.039836518466472626, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5322621464729309, + "rewards/pad": 0.0, + "step": 1216 + }, + { + "completion_length": 123.03125, + "epoch": 0.3878266411727215, + "grad_norm": 14.58663272857666, + "kl": 0.134765625, + "learning_rate": 6.121733588272785e-07, + "loss": 0.0054, + "reward": 1.5366692543029785, + "reward_std": 0.07987186312675476, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.42729437351226807, + "step": 1217 + }, + { + "completion_length": 98.171875, + "epoch": 0.3881453154875717, + "grad_norm": 76.23131561279297, + "kl": 0.166015625, + "learning_rate": 6.118546845124283e-07, + "loss": 0.0066, + "reward": 1.5804510116577148, + "reward_std": 0.15513239800930023, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.47107604146003723, + "rewards/pad": 0.125, + "step": 1218 + }, + { + "completion_length": 125.625, + "epoch": 0.3884639898024219, + "grad_norm": 9.716293334960938, + "kl": 0.166015625, + "learning_rate": 6.11536010197578e-07, + "loss": 0.0066, + "reward": 1.4332759380340576, + "reward_std": 0.043724529445171356, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4332759380340576, + "step": 1219 + }, + { + "completion_length": 174.078125, + "epoch": 0.38878266411727214, + "grad_norm": 5.577003479003906, + "kl": 0.1025390625, + "learning_rate": 6.112173358827278e-07, + "loss": 0.0041, + "reward": 1.3279743194580078, + "reward_std": 0.059924960136413574, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3279743492603302, + "step": 1220 + }, + { + "completion_length": 147.234375, + "epoch": 0.38910133843212236, + "grad_norm": 22.81951904296875, + "kl": 0.10205078125, + "learning_rate": 6.108986615678776e-07, + "loss": 0.0041, + "reward": 1.5294336080551147, + "reward_std": 0.06180447340011597, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.52943354845047, + "step": 1221 + }, + { + "completion_length": 123.5, + "epoch": 0.3894200127469726, + "grad_norm": 17.01264190673828, + "kl": 0.21484375, + "learning_rate": 6.105799872530274e-07, + "loss": 0.0086, + "reward": 1.5035171508789062, + "reward_std": 0.09979550540447235, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5035170912742615, + "rewards/pad": 0.0, + "step": 1222 + }, + { + "completion_length": 73.984375, + "epoch": 0.3897386870618228, + "grad_norm": 13.66805362701416, + "kl": 0.1025390625, + "learning_rate": 6.102613129381771e-07, + "loss": 0.0041, + "reward": 1.6973612308502197, + "reward_std": 0.10417403280735016, + "rewards/answer_reward": 0.359375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.3379861116409302, + "step": 1223 + }, + { + "completion_length": 97.625, + "epoch": 0.390057361376673, + "grad_norm": 45.69769287109375, + "kl": 0.119140625, + "learning_rate": 6.09942638623327e-07, + "loss": 0.0048, + "reward": 1.6858172416687012, + "reward_std": 0.11278460919857025, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.45144230127334595, + "step": 1224 + }, + { + "completion_length": 96.609375, + "epoch": 0.39037603569152324, + "grad_norm": 67.78718566894531, + "kl": 0.1259765625, + "learning_rate": 6.096239643084768e-07, + "loss": 0.005, + "reward": 1.3379242420196533, + "reward_std": 0.05427855625748634, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3379242718219757, + "rewards/pad": 0.0, + "step": 1225 + }, + { + "completion_length": 97.15625, + "epoch": 0.3906947100063735, + "grad_norm": 29.57921600341797, + "kl": 0.1123046875, + "learning_rate": 6.093052899936266e-07, + "loss": 0.0045, + "reward": 1.5308830738067627, + "reward_std": 0.06973686814308167, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5308831334114075, + "rewards/pad": 0.0, + "step": 1226 + }, + { + "completion_length": 97.53125, + "epoch": 0.39101338432122373, + "grad_norm": 38.909488677978516, + "kl": 0.13671875, + "learning_rate": 6.089866156787763e-07, + "loss": 0.0055, + "reward": 1.3697888851165771, + "reward_std": 0.058314502239227295, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3697889447212219, + "step": 1227 + }, + { + "completion_length": 47.421875, + "epoch": 0.39133205863607395, + "grad_norm": 23.21940803527832, + "kl": 0.173828125, + "learning_rate": 6.086679413639261e-07, + "loss": 0.007, + "reward": 1.8355824947357178, + "reward_std": 0.10905968397855759, + "rewards/answer_reward": 0.375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.46058255434036255, + "step": 1228 + }, + { + "completion_length": 145.453125, + "epoch": 0.3916507329509242, + "grad_norm": 33.876712799072266, + "kl": 0.1953125, + "learning_rate": 6.083492670490759e-07, + "loss": 0.0078, + "reward": 1.4986937046051025, + "reward_std": 0.045384809374809265, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.37369370460510254, + "step": 1229 + }, + { + "completion_length": 45.65625, + "epoch": 0.3919694072657744, + "grad_norm": 30.74664878845215, + "kl": 0.1494140625, + "learning_rate": 6.080305927342257e-07, + "loss": 0.006, + "reward": 1.5915486812591553, + "reward_std": 0.09160721302032471, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4665486812591553, + "rewards/pad": 0.125, + "step": 1230 + }, + { + "completion_length": 98.203125, + "epoch": 0.3922880815806246, + "grad_norm": 296.3717346191406, + "kl": 0.259765625, + "learning_rate": 6.077119184193754e-07, + "loss": 0.0104, + "reward": 1.7653553485870361, + "reward_std": 0.21209806203842163, + "rewards/pad": 0.140625, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.6247303485870361, + "step": 1231 + }, + { + "completion_length": 121.4375, + "epoch": 0.39260675589547483, + "grad_norm": 25.76085090637207, + "kl": 0.1689453125, + "learning_rate": 6.073932441045252e-07, + "loss": 0.0068, + "reward": 1.4098402261734009, + "reward_std": 0.061084769666194916, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4098402261734009, + "rewards/pad": 0.0, + "step": 1232 + }, + { + "completion_length": 71.28125, + "epoch": 0.39292543021032506, + "grad_norm": 26.772653579711914, + "kl": 0.189453125, + "learning_rate": 6.07074569789675e-07, + "loss": 0.0076, + "reward": 1.5265135765075684, + "reward_std": 0.057546839118003845, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5265136957168579, + "rewards/pad": 0.0, + "step": 1233 + }, + { + "completion_length": 97.0625, + "epoch": 0.3932441045251753, + "grad_norm": 8.714112281799316, + "kl": 0.1962890625, + "learning_rate": 6.067558954748247e-07, + "loss": 0.0079, + "reward": 1.6967378854751587, + "reward_std": 0.06316147744655609, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5717379450798035, + "step": 1234 + }, + { + "completion_length": 123.671875, + "epoch": 0.3935627788400255, + "grad_norm": 14.036128997802734, + "kl": 0.423828125, + "learning_rate": 6.064372211599745e-07, + "loss": 0.017, + "reward": 1.5465986728668213, + "reward_std": 0.10370682924985886, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5465986132621765, + "step": 1235 + }, + { + "completion_length": 148.5, + "epoch": 0.3938814531548757, + "grad_norm": 7.814291000366211, + "kl": 0.1025390625, + "learning_rate": 6.061185468451242e-07, + "loss": 0.0041, + "reward": 1.4367327690124512, + "reward_std": 0.06396310031414032, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.31173282861709595, + "step": 1236 + }, + { + "completion_length": 95.34375, + "epoch": 0.39420012746972594, + "grad_norm": 16.77633285522461, + "kl": 0.1474609375, + "learning_rate": 6.05799872530274e-07, + "loss": 0.0059, + "reward": 1.5735642910003662, + "reward_std": 0.05050143599510193, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4485642611980438, + "step": 1237 + }, + { + "completion_length": 70.765625, + "epoch": 0.39451880178457616, + "grad_norm": 22.550230026245117, + "kl": 0.15625, + "learning_rate": 6.054811982154237e-07, + "loss": 0.0062, + "reward": 1.5160186290740967, + "reward_std": 0.061520084738731384, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.39101865887641907, + "step": 1238 + }, + { + "completion_length": 95.8125, + "epoch": 0.3948374760994264, + "grad_norm": 58.803009033203125, + "kl": 0.1376953125, + "learning_rate": 6.051625239005735e-07, + "loss": 0.0055, + "reward": 1.5313327312469482, + "reward_std": 0.10405335575342178, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.42195767164230347, + "rewards/pad": 0.109375, + "step": 1239 + }, + { + "completion_length": 122.25, + "epoch": 0.3951561504142766, + "grad_norm": 49.10667037963867, + "kl": 0.455078125, + "learning_rate": 6.048438495857233e-07, + "loss": 0.0182, + "reward": 1.455152988433838, + "reward_std": 0.04565514251589775, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4551529586315155, + "rewards/pad": 0.0, + "step": 1240 + }, + { + "completion_length": 74.921875, + "epoch": 0.3954748247291268, + "grad_norm": 109.96369171142578, + "kl": 0.1875, + "learning_rate": 6.045251752708731e-07, + "loss": 0.0075, + "reward": 1.6491847038269043, + "reward_std": 0.1602613925933838, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5241846442222595, + "rewards/pad": 0.125, + "step": 1241 + }, + { + "completion_length": 72.859375, + "epoch": 0.39579349904397704, + "grad_norm": 14.490010261535645, + "kl": 0.220703125, + "learning_rate": 6.042065009560228e-07, + "loss": 0.0089, + "reward": 1.7809354066848755, + "reward_std": 0.07316883653402328, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5309354066848755, + "step": 1242 + }, + { + "completion_length": 151.890625, + "epoch": 0.39611217335882726, + "grad_norm": 13.692108154296875, + "kl": 0.09326171875, + "learning_rate": 6.038878266411726e-07, + "loss": 0.0037, + "reward": 1.3232471942901611, + "reward_std": 0.04393507167696953, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.32324713468551636, + "rewards/pad": 0.0, + "step": 1243 + }, + { + "completion_length": 146.484375, + "epoch": 0.3964308476736775, + "grad_norm": 12.829249382019043, + "kl": 0.1123046875, + "learning_rate": 6.035691523263225e-07, + "loss": 0.0045, + "reward": 1.5288894176483154, + "reward_std": 0.0912233367562294, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.41951435804367065, + "step": 1244 + }, + { + "completion_length": 123.625, + "epoch": 0.3967495219885277, + "grad_norm": 26.55255126953125, + "kl": 0.10888671875, + "learning_rate": 6.032504780114723e-07, + "loss": 0.0044, + "reward": 1.429291844367981, + "reward_std": 0.05211643874645233, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4292917847633362, + "step": 1245 + }, + { + "completion_length": 71.390625, + "epoch": 0.397068196303378, + "grad_norm": 36.959205627441406, + "kl": 0.22265625, + "learning_rate": 6.02931803696622e-07, + "loss": 0.0089, + "reward": 1.6715713739395142, + "reward_std": 0.05109931528568268, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5465713739395142, + "rewards/pad": 0.125, + "step": 1246 + }, + { + "completion_length": 94.53125, + "epoch": 0.3973868706182282, + "grad_norm": 51.179443359375, + "kl": 0.11083984375, + "learning_rate": 6.026131293817718e-07, + "loss": 0.0044, + "reward": 1.4588894844055176, + "reward_std": 0.10032698512077332, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4588894844055176, + "step": 1247 + }, + { + "completion_length": 69.46875, + "epoch": 0.3977055449330784, + "grad_norm": 101.04873657226562, + "kl": 0.322265625, + "learning_rate": 6.022944550669216e-07, + "loss": 0.0129, + "reward": 1.4367910623550415, + "reward_std": 0.06670445203781128, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3117910623550415, + "rewards/pad": 0.125, + "step": 1248 + }, + { + "completion_length": 48.53125, + "epoch": 0.39802421924792863, + "grad_norm": 58.24769592285156, + "kl": 0.158203125, + "learning_rate": 6.019757807520714e-07, + "loss": 0.0063, + "reward": 1.781355619430542, + "reward_std": 0.20801198482513428, + "rewards/answer_reward": 0.21875, + "rewards/format_reward_gqa": 0.984375, + "rewards/iou_glue_reward": 0.5782306790351868, + "step": 1249 + }, + { + "completion_length": 170.09375, + "epoch": 0.39834289356277885, + "grad_norm": 42.81908416748047, + "kl": 0.0927734375, + "learning_rate": 6.016571064372211e-07, + "loss": 0.0037, + "reward": 1.4066224098205566, + "reward_std": 0.09949782490730286, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.42224735021591187, + "step": 1250 + }, + { + "completion_length": 120.28125, + "epoch": 0.3986615678776291, + "grad_norm": 17.177026748657227, + "kl": 0.11279296875, + "learning_rate": 6.013384321223709e-07, + "loss": 0.0045, + "reward": 1.5096888542175293, + "reward_std": 0.04415253549814224, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5096888542175293, + "rewards/pad": 0.0, + "step": 1251 + }, + { + "completion_length": 119.890625, + "epoch": 0.3989802421924793, + "grad_norm": 31.22040557861328, + "kl": 0.130859375, + "learning_rate": 6.010197578075207e-07, + "loss": 0.0052, + "reward": 1.4680066108703613, + "reward_std": 0.08115118741989136, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4680064916610718, + "step": 1252 + }, + { + "completion_length": 71.28125, + "epoch": 0.3992989165073295, + "grad_norm": 209.50657653808594, + "kl": 0.1376953125, + "learning_rate": 6.007010834926705e-07, + "loss": 0.0055, + "reward": 1.8249537944793701, + "reward_std": 0.1311221420764923, + "rewards/pad": 0.21875, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.6062036752700806, + "step": 1253 + }, + { + "completion_length": 171.140625, + "epoch": 0.39961759082217974, + "grad_norm": 31.1502685546875, + "kl": 0.0927734375, + "learning_rate": 6.003824091778202e-07, + "loss": 0.0037, + "reward": 1.5441827774047852, + "reward_std": 0.05834697559475899, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5441827774047852, + "step": 1254 + }, + { + "completion_length": 172.640625, + "epoch": 0.39993626513702996, + "grad_norm": 13.332192420959473, + "kl": 0.10595703125, + "learning_rate": 6.0006373486297e-07, + "loss": 0.0042, + "reward": 1.4143109321594238, + "reward_std": 0.08136802166700363, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.414311021566391, + "step": 1255 + }, + { + "completion_length": 95.46875, + "epoch": 0.4002549394518802, + "grad_norm": 11.769364356994629, + "kl": 0.154296875, + "learning_rate": 5.997450605481198e-07, + "loss": 0.0062, + "reward": 1.5249242782592773, + "reward_std": 0.10501822829246521, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5249243974685669, + "step": 1256 + }, + { + "completion_length": 146.171875, + "epoch": 0.4005736137667304, + "grad_norm": 40.824153900146484, + "kl": 0.1357421875, + "learning_rate": 5.994263862332696e-07, + "loss": 0.0054, + "reward": 1.3846766948699951, + "reward_std": 0.071880042552948, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.38467681407928467, + "step": 1257 + }, + { + "completion_length": 97.21875, + "epoch": 0.4008922880815806, + "grad_norm": 1223.993408203125, + "kl": 0.103515625, + "learning_rate": 5.991077119184193e-07, + "loss": 0.0041, + "reward": 1.598623514175415, + "reward_std": 0.06979230791330338, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.47362351417541504, + "step": 1258 + }, + { + "completion_length": 94.0625, + "epoch": 0.40121096239643084, + "grad_norm": 30.500272750854492, + "kl": 0.166015625, + "learning_rate": 5.987890376035691e-07, + "loss": 0.0066, + "reward": 1.559830665588379, + "reward_std": 0.07134728133678436, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5598306059837341, + "rewards/pad": 0.0, + "step": 1259 + }, + { + "completion_length": 95.609375, + "epoch": 0.40152963671128106, + "grad_norm": 43.06867218017578, + "kl": 0.140625, + "learning_rate": 5.984703632887189e-07, + "loss": 0.0056, + "reward": 1.3730461597442627, + "reward_std": 0.10210265219211578, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.3886711001396179, + "rewards/pad": 0.0, + "step": 1260 + }, + { + "completion_length": 98.140625, + "epoch": 0.4018483110261313, + "grad_norm": 33.724788665771484, + "kl": 0.16015625, + "learning_rate": 5.981516889738687e-07, + "loss": 0.0064, + "reward": 1.7647151947021484, + "reward_std": 0.06048569828271866, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5147151350975037, + "step": 1261 + }, + { + "completion_length": 73.9375, + "epoch": 0.4021669853409815, + "grad_norm": 37.967552185058594, + "kl": 0.09765625, + "learning_rate": 5.978330146590184e-07, + "loss": 0.0039, + "reward": 1.8287855386734009, + "reward_std": 0.0900324210524559, + "rewards/pad": 0.375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4537854492664337, + "step": 1262 + }, + { + "completion_length": 147.96875, + "epoch": 0.4024856596558317, + "grad_norm": 34.182621002197266, + "kl": 0.0947265625, + "learning_rate": 5.975143403441683e-07, + "loss": 0.0038, + "reward": 1.5652196407318115, + "reward_std": 0.05024642124772072, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.565219521522522, + "step": 1263 + }, + { + "completion_length": 68.875, + "epoch": 0.40280433397068194, + "grad_norm": 82.22328186035156, + "kl": 0.255859375, + "learning_rate": 5.971956660293181e-07, + "loss": 0.0102, + "reward": 1.5273429155349731, + "reward_std": 0.09247814118862152, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.40234294533729553, + "rewards/pad": 0.125, + "step": 1264 + }, + { + "completion_length": 121.5625, + "epoch": 0.40312300828553216, + "grad_norm": 46.81608200073242, + "kl": 0.1201171875, + "learning_rate": 5.968769917144678e-07, + "loss": 0.0048, + "reward": 1.4116365909576416, + "reward_std": 0.12033785134553909, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.427261620759964, + "step": 1265 + }, + { + "completion_length": 92.359375, + "epoch": 0.40344168260038243, + "grad_norm": 35.47514343261719, + "kl": 0.1318359375, + "learning_rate": 5.965583173996176e-07, + "loss": 0.0053, + "reward": 1.4215552806854248, + "reward_std": 0.036197151988744736, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.42155537009239197, + "step": 1266 + }, + { + "completion_length": 95.3125, + "epoch": 0.40376035691523265, + "grad_norm": 51.23936080932617, + "kl": 0.1787109375, + "learning_rate": 5.962396430847674e-07, + "loss": 0.0072, + "reward": 1.5264898538589478, + "reward_std": 0.1344299465417862, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.43273985385894775, + "rewards/pad": 0.09375, + "step": 1267 + }, + { + "completion_length": 70.84375, + "epoch": 0.4040790312300829, + "grad_norm": 33.75147247314453, + "kl": 0.158203125, + "learning_rate": 5.959209687699172e-07, + "loss": 0.0063, + "reward": 1.6157795190811157, + "reward_std": 0.054706256836652756, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.6157793998718262, + "step": 1268 + }, + { + "completion_length": 71.5, + "epoch": 0.4043977055449331, + "grad_norm": 75.2858657836914, + "kl": 0.1533203125, + "learning_rate": 5.956022944550669e-07, + "loss": 0.0061, + "reward": 1.7125145196914673, + "reward_std": 0.11750952899456024, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4625145196914673, + "rewards/pad": 0.25, + "step": 1269 + }, + { + "completion_length": 146.171875, + "epoch": 0.4047163798597833, + "grad_norm": 46.43769454956055, + "kl": 0.0888671875, + "learning_rate": 5.952836201402167e-07, + "loss": 0.0035, + "reward": 1.4164249897003174, + "reward_std": 0.1052272617816925, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.41642504930496216, + "rewards/pad": 0.0, + "step": 1270 + }, + { + "completion_length": 118.015625, + "epoch": 0.40503505417463354, + "grad_norm": 20.825685501098633, + "kl": 0.2333984375, + "learning_rate": 5.949649458253665e-07, + "loss": 0.0093, + "reward": 1.519181251525879, + "reward_std": 0.062050338834524155, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5191812515258789, + "step": 1271 + }, + { + "completion_length": 45.78125, + "epoch": 0.40535372848948376, + "grad_norm": 35.329437255859375, + "kl": 0.21875, + "learning_rate": 5.946462715105163e-07, + "loss": 0.0088, + "reward": 1.4968239068984985, + "reward_std": 0.129884272813797, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.49682390689849854, + "rewards/pad": 0.0, + "step": 1272 + }, + { + "completion_length": 19.328125, + "epoch": 0.405672402804334, + "grad_norm": 82.19258880615234, + "kl": 0.203125, + "learning_rate": 5.94327597195666e-07, + "loss": 0.0082, + "reward": 1.5742777585983276, + "reward_std": 0.10323308408260345, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5742777585983276, + "rewards/pad": 0.0, + "step": 1273 + }, + { + "completion_length": 94.3125, + "epoch": 0.4059910771191842, + "grad_norm": 27.67733383178711, + "kl": 0.1455078125, + "learning_rate": 5.940089228808158e-07, + "loss": 0.0058, + "reward": 1.681000828742981, + "reward_std": 0.06062120571732521, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.556000828742981, + "rewards/pad": 0.125, + "step": 1274 + }, + { + "completion_length": 96.765625, + "epoch": 0.4063097514340344, + "grad_norm": 36.68671417236328, + "kl": 0.134765625, + "learning_rate": 5.936902485659655e-07, + "loss": 0.0054, + "reward": 1.4792072772979736, + "reward_std": 0.053634293377399445, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.47920721769332886, + "step": 1275 + }, + { + "completion_length": 147.859375, + "epoch": 0.40662842574888464, + "grad_norm": 7.783206939697266, + "kl": 0.111328125, + "learning_rate": 5.933715742511153e-07, + "loss": 0.0045, + "reward": 1.4867475032806396, + "reward_std": 0.04772502928972244, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4867474436759949, + "step": 1276 + }, + { + "completion_length": 43.28125, + "epoch": 0.40694710006373486, + "grad_norm": 40.26476287841797, + "kl": 0.1953125, + "learning_rate": 5.93052899936265e-07, + "loss": 0.0078, + "reward": 1.5758988857269287, + "reward_std": 0.0841723382472992, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5758988857269287, + "rewards/pad": 0.0, + "step": 1277 + }, + { + "completion_length": 93.84375, + "epoch": 0.4072657743785851, + "grad_norm": 20.492387771606445, + "kl": 0.1298828125, + "learning_rate": 5.927342256214148e-07, + "loss": 0.0052, + "reward": 1.5350046157836914, + "reward_std": 0.027536258101463318, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.410004585981369, + "rewards/pad": 0.125, + "step": 1278 + }, + { + "completion_length": 94.640625, + "epoch": 0.4075844486934353, + "grad_norm": 15.450780868530273, + "kl": 0.146484375, + "learning_rate": 5.924155513065646e-07, + "loss": 0.0059, + "reward": 1.612504482269287, + "reward_std": 0.09879954159259796, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6125044822692871, + "rewards/pad": 0.0, + "step": 1279 + }, + { + "completion_length": 69.890625, + "epoch": 0.4079031230082855, + "grad_norm": 16.371076583862305, + "kl": 0.2373046875, + "learning_rate": 5.920968769917144e-07, + "loss": 0.0095, + "reward": 1.6929583549499512, + "reward_std": 0.037510331720113754, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5679583549499512, + "rewards/pad": 0.125, + "step": 1280 + }, + { + "completion_length": 95.890625, + "epoch": 0.40822179732313574, + "grad_norm": 33.37834548950195, + "kl": 0.11865234375, + "learning_rate": 5.917782026768641e-07, + "loss": 0.0048, + "reward": 1.5515503883361816, + "reward_std": 0.085669606924057, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5515503883361816, + "rewards/pad": 0.0, + "step": 1281 + }, + { + "completion_length": 45.03125, + "epoch": 0.40854047163798596, + "grad_norm": 67.09070587158203, + "kl": 0.1513671875, + "learning_rate": 5.91459528362014e-07, + "loss": 0.0061, + "reward": 1.7493417263031006, + "reward_std": 0.11784421652555466, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.624341607093811, + "rewards/pad": 0.125, + "step": 1282 + }, + { + "completion_length": 120.34375, + "epoch": 0.4088591459528362, + "grad_norm": 125.51654815673828, + "kl": 0.1533203125, + "learning_rate": 5.911408540471638e-07, + "loss": 0.0061, + "reward": 1.40920090675354, + "reward_std": 0.09400975704193115, + "rewards/pad": 0.03125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3779507875442505, + "step": 1283 + }, + { + "completion_length": 69.625, + "epoch": 0.4091778202676864, + "grad_norm": 33.30241394042969, + "kl": 0.1298828125, + "learning_rate": 5.908221797323136e-07, + "loss": 0.0052, + "reward": 1.46676766872406, + "reward_std": 0.05991966277360916, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.34176772832870483, + "rewards/pad": 0.125, + "step": 1284 + }, + { + "completion_length": 146.0625, + "epoch": 0.4094964945825367, + "grad_norm": 25.540910720825195, + "kl": 0.08740234375, + "learning_rate": 5.905035054174633e-07, + "loss": 0.0035, + "reward": 1.4206722974777222, + "reward_std": 0.1471319943666458, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.38942232728004456, + "rewards/pad": 0.046875, + "step": 1285 + }, + { + "completion_length": 155.65625, + "epoch": 0.4098151688973869, + "grad_norm": 117.83950805664062, + "kl": 0.08544921875, + "learning_rate": 5.901848311026131e-07, + "loss": 0.0034, + "reward": 1.3239874839782715, + "reward_std": 0.13704180717468262, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.33961254358291626, + "step": 1286 + }, + { + "completion_length": 95.28125, + "epoch": 0.4101338432122371, + "grad_norm": 34.79705047607422, + "kl": 0.1376953125, + "learning_rate": 5.898661567877629e-07, + "loss": 0.0055, + "reward": 1.5317813158035278, + "reward_std": 0.11902156472206116, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4692813456058502, + "rewards/pad": 0.0625, + "step": 1287 + }, + { + "completion_length": 147.984375, + "epoch": 0.41045251752708733, + "grad_norm": 44.45170211791992, + "kl": 0.07470703125, + "learning_rate": 5.895474824729127e-07, + "loss": 0.003, + "reward": 1.4936556816101074, + "reward_std": 0.05926290154457092, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.49365559220314026, + "step": 1288 + }, + { + "completion_length": 120.65625, + "epoch": 0.41077119184193756, + "grad_norm": 21.58782196044922, + "kl": 0.10205078125, + "learning_rate": 5.892288081580624e-07, + "loss": 0.0041, + "reward": 1.623151183128357, + "reward_std": 0.07410305738449097, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.49815118312835693, + "step": 1289 + }, + { + "completion_length": 123.046875, + "epoch": 0.4110898661567878, + "grad_norm": 23.982749938964844, + "kl": 0.09716796875, + "learning_rate": 5.889101338432122e-07, + "loss": 0.0039, + "reward": 1.4851818084716797, + "reward_std": 0.11753320693969727, + "rewards/pad": 0.03125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.45393168926239014, + "step": 1290 + }, + { + "completion_length": 120.84375, + "epoch": 0.411408540471638, + "grad_norm": 12.221750259399414, + "kl": 0.1318359375, + "learning_rate": 5.88591459528362e-07, + "loss": 0.0053, + "reward": 1.3529877662658691, + "reward_std": 0.028966424986720085, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.35298776626586914, + "rewards/pad": 0.0, + "step": 1291 + }, + { + "completion_length": 95.75, + "epoch": 0.4117272147864882, + "grad_norm": 87.19110870361328, + "kl": 0.15625, + "learning_rate": 5.882727852135117e-07, + "loss": 0.0062, + "reward": 1.5021991729736328, + "reward_std": 0.04907491058111191, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3771991431713104, + "rewards/pad": 0.125, + "step": 1292 + }, + { + "completion_length": 47.515625, + "epoch": 0.41204588910133844, + "grad_norm": 43.21669387817383, + "kl": 0.150390625, + "learning_rate": 5.879541108986615e-07, + "loss": 0.006, + "reward": 1.508324146270752, + "reward_std": 0.05069003999233246, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.38332417607307434, + "step": 1293 + }, + { + "completion_length": 93.15625, + "epoch": 0.41236456341618866, + "grad_norm": 60.25608444213867, + "kl": 0.12255859375, + "learning_rate": 5.876354365838113e-07, + "loss": 0.0049, + "reward": 1.5370088815689087, + "reward_std": 0.0394783578813076, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5370088219642639, + "step": 1294 + }, + { + "completion_length": 94.59375, + "epoch": 0.4126832377310389, + "grad_norm": 23.95777130126953, + "kl": 0.11376953125, + "learning_rate": 5.873167622689611e-07, + "loss": 0.0046, + "reward": 1.524670124053955, + "reward_std": 0.08892560005187988, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3996701240539551, + "rewards/pad": 0.125, + "step": 1295 + }, + { + "completion_length": 119.328125, + "epoch": 0.4130019120458891, + "grad_norm": 22.234315872192383, + "kl": 0.0810546875, + "learning_rate": 5.869980879541108e-07, + "loss": 0.0033, + "reward": 1.7183890342712402, + "reward_std": 0.06814469397068024, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.48401397466659546, + "rewards/pad": 0.234375, + "step": 1296 + }, + { + "completion_length": 45.359375, + "epoch": 0.4133205863607393, + "grad_norm": 27.19536018371582, + "kl": 0.330078125, + "learning_rate": 5.866794136392606e-07, + "loss": 0.0132, + "reward": 1.650060772895813, + "reward_std": 0.1549336016178131, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.47818565368652344, + "rewards/pad": 0.171875, + "step": 1297 + }, + { + "completion_length": 120.390625, + "epoch": 0.41363926067558954, + "grad_norm": 662.822998046875, + "kl": 0.14453125, + "learning_rate": 5.863607393244104e-07, + "loss": 0.0058, + "reward": 1.4083480834960938, + "reward_std": 0.07196545600891113, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.408348023891449, + "rewards/pad": 0.0, + "step": 1298 + }, + { + "completion_length": 94.171875, + "epoch": 0.41395793499043976, + "grad_norm": 29.26519775390625, + "kl": 0.1376953125, + "learning_rate": 5.860420650095602e-07, + "loss": 0.0055, + "reward": 1.5837488174438477, + "reward_std": 0.07130052894353867, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.45874881744384766, + "step": 1299 + }, + { + "completion_length": 70.515625, + "epoch": 0.41427660930529, + "grad_norm": 48.05635070800781, + "kl": 0.1982421875, + "learning_rate": 5.8572339069471e-07, + "loss": 0.0079, + "reward": 1.6558793783187866, + "reward_std": 0.27641844749450684, + "rewards/pad": 0.171875, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.49962949752807617, + "step": 1300 + }, + { + "completion_length": 68.953125, + "epoch": 0.4145952836201402, + "grad_norm": 28.768461227416992, + "kl": 0.1279296875, + "learning_rate": 5.854047163798598e-07, + "loss": 0.0051, + "reward": 1.7155733108520508, + "reward_std": 0.05253777652978897, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5905733704566956, + "rewards/pad": 0.125, + "step": 1301 + }, + { + "completion_length": 69.15625, + "epoch": 0.4149139579349904, + "grad_norm": 130.46864318847656, + "kl": 0.11962890625, + "learning_rate": 5.850860420650096e-07, + "loss": 0.0048, + "reward": 1.496382474899292, + "reward_std": 0.11643992364406586, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.371382474899292, + "rewards/pad": 0.125, + "step": 1302 + }, + { + "completion_length": 94.84375, + "epoch": 0.41523263224984064, + "grad_norm": 20.442569732666016, + "kl": 0.12060546875, + "learning_rate": 5.847673677501594e-07, + "loss": 0.0048, + "reward": 1.4754797220230103, + "reward_std": 0.07423287630081177, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.47547969222068787, + "rewards/pad": 0.0, + "step": 1303 + }, + { + "completion_length": 146.1875, + "epoch": 0.41555130656469086, + "grad_norm": 45.905033111572266, + "kl": 0.142578125, + "learning_rate": 5.844486934353091e-07, + "loss": 0.0057, + "reward": 1.283578872680664, + "reward_std": 0.0793718472123146, + "rewards/pad": 0.015625, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.26795390248298645, + "step": 1304 + }, + { + "completion_length": 92.78125, + "epoch": 0.41586998087954113, + "grad_norm": 50.72040939331055, + "kl": 0.1884765625, + "learning_rate": 5.841300191204589e-07, + "loss": 0.0075, + "reward": 1.4547278881072998, + "reward_std": 0.069220632314682, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4547279179096222, + "step": 1305 + }, + { + "completion_length": 69.046875, + "epoch": 0.41618865519439135, + "grad_norm": 38.16029357910156, + "kl": 0.142578125, + "learning_rate": 5.838113448056087e-07, + "loss": 0.0057, + "reward": 1.5830930471420288, + "reward_std": 0.05439729988574982, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5830931663513184, + "step": 1306 + }, + { + "completion_length": 170.390625, + "epoch": 0.4165073295092416, + "grad_norm": 27.359622955322266, + "kl": 0.07861328125, + "learning_rate": 5.834926704907585e-07, + "loss": 0.0032, + "reward": 1.576507329940796, + "reward_std": 0.051485076546669006, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5765073299407959, + "step": 1307 + }, + { + "completion_length": 144.578125, + "epoch": 0.4168260038240918, + "grad_norm": 18.980205535888672, + "kl": 0.1669921875, + "learning_rate": 5.831739961759082e-07, + "loss": 0.0067, + "reward": 1.5791575908660889, + "reward_std": 0.04680941626429558, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.45415759086608887, + "rewards/pad": 0.125, + "step": 1308 + }, + { + "completion_length": 96.25, + "epoch": 0.417144678138942, + "grad_norm": 38.10404968261719, + "kl": 0.224609375, + "learning_rate": 5.82855321861058e-07, + "loss": 0.009, + "reward": 1.595242977142334, + "reward_std": 0.07397254556417465, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.34524303674697876, + "step": 1309 + }, + { + "completion_length": 70.515625, + "epoch": 0.41746335245379224, + "grad_norm": 18.30024528503418, + "kl": 0.1884765625, + "learning_rate": 5.825366475462078e-07, + "loss": 0.0075, + "reward": 1.6927845478057861, + "reward_std": 0.06641843169927597, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.44278454780578613, + "step": 1310 + }, + { + "completion_length": 147.203125, + "epoch": 0.41778202676864246, + "grad_norm": 18.82292366027832, + "kl": 0.08935546875, + "learning_rate": 5.822179732313576e-07, + "loss": 0.0036, + "reward": 1.5743370056152344, + "reward_std": 0.05167382210493088, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4493370056152344, + "step": 1311 + }, + { + "completion_length": 71.265625, + "epoch": 0.4181007010834927, + "grad_norm": 36.481136322021484, + "kl": 0.1396484375, + "learning_rate": 5.818992989165073e-07, + "loss": 0.0056, + "reward": 1.5864616632461548, + "reward_std": 0.06692761927843094, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4614616930484772, + "rewards/pad": 0.125, + "step": 1312 + }, + { + "completion_length": 44.734375, + "epoch": 0.4184193753983429, + "grad_norm": 26.760005950927734, + "kl": 0.1552734375, + "learning_rate": 5.815806246016571e-07, + "loss": 0.0062, + "reward": 1.8370531797409058, + "reward_std": 0.05633014068007469, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5870531797409058, + "rewards/pad": 0.25, + "step": 1313 + }, + { + "completion_length": 174.125, + "epoch": 0.4187380497131931, + "grad_norm": 17.700157165527344, + "kl": 0.064453125, + "learning_rate": 5.812619502868069e-07, + "loss": 0.0026, + "reward": 1.5748400688171387, + "reward_std": 0.11043757945299149, + "rewards/pad": 0.078125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.49671506881713867, + "step": 1314 + }, + { + "completion_length": 122.75, + "epoch": 0.41905672402804334, + "grad_norm": 112.10523986816406, + "kl": 0.1201171875, + "learning_rate": 5.809432759719566e-07, + "loss": 0.0048, + "reward": 1.4939663410186768, + "reward_std": 0.07092640548944473, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.493966281414032, + "step": 1315 + }, + { + "completion_length": 93.890625, + "epoch": 0.41937539834289356, + "grad_norm": 68.99323272705078, + "kl": 0.1396484375, + "learning_rate": 5.806246016571063e-07, + "loss": 0.0056, + "reward": 1.399767518043518, + "reward_std": 0.05090294033288956, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.39976757764816284, + "rewards/pad": 0.0, + "step": 1316 + }, + { + "completion_length": 97.109375, + "epoch": 0.4196940726577438, + "grad_norm": 36.900672912597656, + "kl": 0.1259765625, + "learning_rate": 5.803059273422561e-07, + "loss": 0.005, + "reward": 1.5363261699676514, + "reward_std": 0.061481181532144547, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.28632619976997375, + "rewards/pad": 0.25, + "step": 1317 + }, + { + "completion_length": 95.265625, + "epoch": 0.420012746972594, + "grad_norm": 69.90725708007812, + "kl": 0.17578125, + "learning_rate": 5.799872530274059e-07, + "loss": 0.007, + "reward": 1.4442648887634277, + "reward_std": 0.09392991662025452, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.44426485896110535, + "rewards/pad": 0.0, + "step": 1318 + }, + { + "completion_length": 97.703125, + "epoch": 0.4203314212874442, + "grad_norm": 18.30790138244629, + "kl": 0.2216796875, + "learning_rate": 5.796685787125558e-07, + "loss": 0.0089, + "reward": 1.4277915954589844, + "reward_std": 0.13965944945812225, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.396541565656662, + "rewards/pad": 0.03125, + "step": 1319 + }, + { + "completion_length": 121.375, + "epoch": 0.42065009560229444, + "grad_norm": 14.09908676147461, + "kl": 0.13671875, + "learning_rate": 5.793499043977055e-07, + "loss": 0.0055, + "reward": 1.3608624935150146, + "reward_std": 0.054107192903757095, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3608624339103699, + "step": 1320 + }, + { + "completion_length": 44.21875, + "epoch": 0.42096876991714466, + "grad_norm": 18.591712951660156, + "kl": 0.23046875, + "learning_rate": 5.790312300828553e-07, + "loss": 0.0092, + "reward": 1.8102295398712158, + "reward_std": 0.15906178951263428, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.7008545994758606, + "rewards/pad": 0.125, + "step": 1321 + }, + { + "completion_length": 45.78125, + "epoch": 0.4212874442319949, + "grad_norm": 68.0612564086914, + "kl": 0.1455078125, + "learning_rate": 5.787125557680051e-07, + "loss": 0.0058, + "reward": 1.786529541015625, + "reward_std": 0.1605585217475891, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.5521544218063354, + "step": 1322 + }, + { + "completion_length": 98.8125, + "epoch": 0.4216061185468451, + "grad_norm": 37.890174865722656, + "kl": 0.09326171875, + "learning_rate": 5.783938814531548e-07, + "loss": 0.0037, + "reward": 1.6999619007110596, + "reward_std": 0.08607161045074463, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.46558693051338196, + "rewards/pad": 0.25, + "step": 1323 + }, + { + "completion_length": 145.75, + "epoch": 0.4219247928616954, + "grad_norm": 14.982475280761719, + "kl": 0.11962890625, + "learning_rate": 5.780752071383046e-07, + "loss": 0.0048, + "reward": 1.6025621891021729, + "reward_std": 0.12264655530452728, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.6181871294975281, + "rewards/pad": 0.0, + "step": 1324 + }, + { + "completion_length": 69.53125, + "epoch": 0.4222434671765456, + "grad_norm": 10.594503402709961, + "kl": 0.146484375, + "learning_rate": 5.777565328234544e-07, + "loss": 0.0058, + "reward": 1.4348278045654297, + "reward_std": 0.03396230563521385, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4348278343677521, + "rewards/pad": 0.0, + "step": 1325 + }, + { + "completion_length": 45.4375, + "epoch": 0.4225621414913958, + "grad_norm": 28.847732543945312, + "kl": 0.162109375, + "learning_rate": 5.774378585086042e-07, + "loss": 0.0065, + "reward": 1.710758924484253, + "reward_std": 0.166127011179924, + "rewards/format_reward_tg": 0.96875, + "rewards/iou_timestamp_reward": 0.49200892448425293, + "rewards/pad": 0.25, + "step": 1326 + }, + { + "completion_length": 46.65625, + "epoch": 0.42288081580624604, + "grad_norm": 62.12076950073242, + "kl": 0.1669921875, + "learning_rate": 5.771191841937539e-07, + "loss": 0.0067, + "reward": 1.4910268783569336, + "reward_std": 0.08188996464014053, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.49102693796157837, + "rewards/pad": 0.0, + "step": 1327 + }, + { + "completion_length": 121.109375, + "epoch": 0.42319949012109626, + "grad_norm": 27.380247116088867, + "kl": 0.13671875, + "learning_rate": 5.768005098789037e-07, + "loss": 0.0055, + "reward": 1.342002511024475, + "reward_std": 0.07291238009929657, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3420025706291199, + "rewards/pad": 0.0, + "step": 1328 + }, + { + "completion_length": 151.125, + "epoch": 0.4235181644359465, + "grad_norm": 22.531246185302734, + "kl": 0.10888671875, + "learning_rate": 5.764818355640535e-07, + "loss": 0.0044, + "reward": 1.7052443027496338, + "reward_std": 0.08555316925048828, + "rewards/pad": 0.109375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5958691835403442, + "step": 1329 + }, + { + "completion_length": 70.671875, + "epoch": 0.4238368387507967, + "grad_norm": 45.3145751953125, + "kl": 0.23828125, + "learning_rate": 5.761631612492033e-07, + "loss": 0.0095, + "reward": 1.6560629606246948, + "reward_std": 0.1088961511850357, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5310629606246948, + "rewards/pad": 0.125, + "step": 1330 + }, + { + "completion_length": 120.609375, + "epoch": 0.4241555130656469, + "grad_norm": 27.0241641998291, + "kl": 0.1533203125, + "learning_rate": 5.75844486934353e-07, + "loss": 0.0061, + "reward": 1.6354317665100098, + "reward_std": 0.0771084874868393, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6354317665100098, + "rewards/pad": 0.0, + "step": 1331 + }, + { + "completion_length": 95.5, + "epoch": 0.42447418738049714, + "grad_norm": 106.8154525756836, + "kl": 0.16015625, + "learning_rate": 5.755258126195028e-07, + "loss": 0.0064, + "reward": 1.7053183317184448, + "reward_std": 0.10672757029533386, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5803183317184448, + "rewards/pad": 0.125, + "step": 1332 + }, + { + "completion_length": 71.421875, + "epoch": 0.42479286169534736, + "grad_norm": 30.220779418945312, + "kl": 0.1484375, + "learning_rate": 5.752071383046526e-07, + "loss": 0.0059, + "reward": 1.4662553071975708, + "reward_std": 0.12006659805774689, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4662552773952484, + "rewards/pad": 0.0, + "step": 1333 + }, + { + "completion_length": 96.25, + "epoch": 0.4251115360101976, + "grad_norm": 52.073760986328125, + "kl": 0.119140625, + "learning_rate": 5.748884639898024e-07, + "loss": 0.0048, + "reward": 1.6597801446914673, + "reward_std": 0.04236849769949913, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.6597801446914673, + "step": 1334 + }, + { + "completion_length": 126.703125, + "epoch": 0.4254302103250478, + "grad_norm": 28.545778274536133, + "kl": 0.10888671875, + "learning_rate": 5.745697896749521e-07, + "loss": 0.0043, + "reward": 1.6625990867614746, + "reward_std": 0.1483616977930069, + "rewards/pad": 0.109375, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.5688490271568298, + "step": 1335 + }, + { + "completion_length": 71.21875, + "epoch": 0.425748884639898, + "grad_norm": 12.07433795928955, + "kl": 0.1455078125, + "learning_rate": 5.742511153601019e-07, + "loss": 0.0058, + "reward": 1.582622766494751, + "reward_std": 0.05007660388946533, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5826227068901062, + "rewards/pad": 0.0, + "step": 1336 + }, + { + "completion_length": 94.796875, + "epoch": 0.42606755895474824, + "grad_norm": 36.460994720458984, + "kl": 0.12353515625, + "learning_rate": 5.739324410452517e-07, + "loss": 0.0049, + "reward": 1.3425620794296265, + "reward_std": 0.049371980130672455, + "rewards/answer_reward": 0.0, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.34256210923194885, + "step": 1337 + }, + { + "completion_length": 68.984375, + "epoch": 0.42638623326959846, + "grad_norm": 31.095317840576172, + "kl": 0.2001953125, + "learning_rate": 5.736137667304016e-07, + "loss": 0.008, + "reward": 1.4534603357315063, + "reward_std": 0.055513009428977966, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.45346033573150635, + "rewards/pad": 0.0, + "step": 1338 + }, + { + "completion_length": 130.03125, + "epoch": 0.4267049075844487, + "grad_norm": 15.608606338500977, + "kl": 0.11083984375, + "learning_rate": 5.732950924155513e-07, + "loss": 0.0044, + "reward": 1.6918649673461914, + "reward_std": 0.08464628458023071, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.44186487793922424, + "step": 1339 + }, + { + "completion_length": 22.265625, + "epoch": 0.4270235818992989, + "grad_norm": 46.421653747558594, + "kl": 0.1572265625, + "learning_rate": 5.729764181007011e-07, + "loss": 0.0063, + "reward": 1.7511345148086548, + "reward_std": 0.1293444037437439, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.5167595148086548, + "rewards/pad": 0.25, + "step": 1340 + }, + { + "completion_length": 70.453125, + "epoch": 0.4273422562141491, + "grad_norm": 212.90789794921875, + "kl": 0.1650390625, + "learning_rate": 5.726577437858509e-07, + "loss": 0.0066, + "reward": 1.647883653640747, + "reward_std": 0.08742669224739075, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5228837728500366, + "rewards/pad": 0.125, + "step": 1341 + }, + { + "completion_length": 72.703125, + "epoch": 0.42766093052899934, + "grad_norm": 18.851896286010742, + "kl": 0.22265625, + "learning_rate": 5.723390694710007e-07, + "loss": 0.0089, + "reward": 1.5048867464065552, + "reward_std": 0.09711531549692154, + "rewards/answer_reward": 0.03125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4736367166042328, + "step": 1342 + }, + { + "completion_length": 70.75, + "epoch": 0.42797960484384956, + "grad_norm": 170.10140991210938, + "kl": 0.1767578125, + "learning_rate": 5.720203951561504e-07, + "loss": 0.0071, + "reward": 1.5883522033691406, + "reward_std": 0.1104811504483223, + "rewards/pad": 0.015625, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5727271437644958, + "step": 1343 + }, + { + "completion_length": 70.359375, + "epoch": 0.42829827915869984, + "grad_norm": 24.550609588623047, + "kl": 0.2109375, + "learning_rate": 5.717017208413002e-07, + "loss": 0.0084, + "reward": 1.7100211381912231, + "reward_std": 0.14169202744960785, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5850211381912231, + "step": 1344 + }, + { + "completion_length": 95.859375, + "epoch": 0.42861695347355006, + "grad_norm": 28.526103973388672, + "kl": 0.251953125, + "learning_rate": 5.7138304652645e-07, + "loss": 0.0101, + "reward": 1.5001425743103027, + "reward_std": 0.08676139265298843, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.500142514705658, + "step": 1345 + }, + { + "completion_length": 70.109375, + "epoch": 0.4289356277884003, + "grad_norm": 47.61008834838867, + "kl": 0.162109375, + "learning_rate": 5.710643722115998e-07, + "loss": 0.0065, + "reward": 1.60111665725708, + "reward_std": 0.06631819903850555, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6011165380477905, + "rewards/pad": 0.0, + "step": 1346 + }, + { + "completion_length": 98.625, + "epoch": 0.4292543021032505, + "grad_norm": 184.50650024414062, + "kl": 0.11572265625, + "learning_rate": 5.707456978967495e-07, + "loss": 0.0046, + "reward": 1.3712372779846191, + "reward_std": 0.09562046080827713, + "rewards/answer_reward": 0.078125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.2931123971939087, + "step": 1347 + }, + { + "completion_length": 123.265625, + "epoch": 0.4295729764181007, + "grad_norm": 40.26362228393555, + "kl": 0.1533203125, + "learning_rate": 5.704270235818993e-07, + "loss": 0.0061, + "reward": 1.5752763748168945, + "reward_std": 0.11223172396421432, + "rewards/pad": 0.09375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4815264046192169, + "step": 1348 + }, + { + "completion_length": 149.0, + "epoch": 0.42989165073295094, + "grad_norm": 45.832801818847656, + "kl": 0.109375, + "learning_rate": 5.701083492670491e-07, + "loss": 0.0044, + "reward": 1.524371862411499, + "reward_std": 0.06820206344127655, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5243719220161438, + "step": 1349 + }, + { + "completion_length": 96.921875, + "epoch": 0.43021032504780116, + "grad_norm": 24.57037353515625, + "kl": 0.1318359375, + "learning_rate": 5.697896749521989e-07, + "loss": 0.0053, + "reward": 1.579499363899231, + "reward_std": 0.05558115616440773, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.45449933409690857, + "step": 1350 + }, + { + "completion_length": 123.171875, + "epoch": 0.4305289993626514, + "grad_norm": 26.439207077026367, + "kl": 0.1162109375, + "learning_rate": 5.694710006373486e-07, + "loss": 0.0047, + "reward": 1.520458459854126, + "reward_std": 0.04056993126869202, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.520458459854126, + "rewards/pad": 0.0, + "step": 1351 + }, + { + "completion_length": 95.734375, + "epoch": 0.4308476736775016, + "grad_norm": 150.6898651123047, + "kl": 0.1298828125, + "learning_rate": 5.691523263224984e-07, + "loss": 0.0052, + "reward": 1.5843548774719238, + "reward_std": 0.11656299233436584, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.47497978806495667, + "rewards/pad": 0.109375, + "step": 1352 + }, + { + "completion_length": 44.890625, + "epoch": 0.4311663479923518, + "grad_norm": 124.51334381103516, + "kl": 0.1337890625, + "learning_rate": 5.688336520076482e-07, + "loss": 0.0054, + "reward": 1.6478725671768188, + "reward_std": 0.13374654948711395, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6478725671768188, + "rewards/pad": 0.0, + "step": 1353 + }, + { + "completion_length": 71.90625, + "epoch": 0.43148502230720204, + "grad_norm": 27.88372230529785, + "kl": 0.1025390625, + "learning_rate": 5.685149776927978e-07, + "loss": 0.0041, + "reward": 1.6494276523590088, + "reward_std": 0.11932900547981262, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6181777715682983, + "rewards/pad": 0.03125, + "step": 1354 + }, + { + "completion_length": 127.421875, + "epoch": 0.43180369662205226, + "grad_norm": 13.997013092041016, + "kl": 0.12451171875, + "learning_rate": 5.681963033779476e-07, + "loss": 0.005, + "reward": 1.6400054693222046, + "reward_std": 0.05574731528759003, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5150054693222046, + "step": 1355 + }, + { + "completion_length": 69.90625, + "epoch": 0.4321223709369025, + "grad_norm": 59.1829719543457, + "kl": 0.1337890625, + "learning_rate": 5.678776290630974e-07, + "loss": 0.0053, + "reward": 1.4801830053329468, + "reward_std": 0.08168666064739227, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.48018306493759155, + "rewards/pad": 0.0, + "step": 1356 + }, + { + "completion_length": 96.53125, + "epoch": 0.4324410452517527, + "grad_norm": 17.930870056152344, + "kl": 0.1474609375, + "learning_rate": 5.675589547482473e-07, + "loss": 0.0059, + "reward": 1.7791249752044678, + "reward_std": 0.09777411073446274, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5291250348091125, + "rewards/pad": 0.25, + "step": 1357 + }, + { + "completion_length": 72.84375, + "epoch": 0.4327597195666029, + "grad_norm": 15.228046417236328, + "kl": 0.1845703125, + "learning_rate": 5.67240280433397e-07, + "loss": 0.0074, + "reward": 1.8047544956207275, + "reward_std": 0.0883319303393364, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5547544956207275, + "step": 1358 + }, + { + "completion_length": 96.8125, + "epoch": 0.43307839388145314, + "grad_norm": 47.425968170166016, + "kl": 0.09423828125, + "learning_rate": 5.669216061185468e-07, + "loss": 0.0038, + "reward": 1.6819376945495605, + "reward_std": 0.07194778323173523, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5569379329681396, + "step": 1359 + }, + { + "completion_length": 155.671875, + "epoch": 0.43339706819630336, + "grad_norm": 27.573013305664062, + "kl": 0.11376953125, + "learning_rate": 5.666029318036966e-07, + "loss": 0.0045, + "reward": 1.5436911582946777, + "reward_std": 0.06996867060661316, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4186912178993225, + "step": 1360 + }, + { + "completion_length": 69.28125, + "epoch": 0.4337157425111536, + "grad_norm": 34.57221984863281, + "kl": 0.142578125, + "learning_rate": 5.662842574888464e-07, + "loss": 0.0057, + "reward": 1.4558401107788086, + "reward_std": 0.08932884782552719, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4558401107788086, + "rewards/pad": 0.0, + "step": 1361 + }, + { + "completion_length": 97.265625, + "epoch": 0.4340344168260038, + "grad_norm": 34.02722930908203, + "kl": 0.1416015625, + "learning_rate": 5.659655831739961e-07, + "loss": 0.0056, + "reward": 1.4915645122528076, + "reward_std": 0.041257619857788086, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4915645718574524, + "rewards/pad": 0.0, + "step": 1362 + }, + { + "completion_length": 95.59375, + "epoch": 0.434353091140854, + "grad_norm": 75.93492126464844, + "kl": 0.1474609375, + "learning_rate": 5.656469088591459e-07, + "loss": 0.0059, + "reward": 1.7674086093902588, + "reward_std": 0.10432854294776917, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6424086093902588, + "rewards/pad": 0.125, + "step": 1363 + }, + { + "completion_length": 71.265625, + "epoch": 0.4346717654557043, + "grad_norm": 31.960256576538086, + "kl": 0.166015625, + "learning_rate": 5.653282345442957e-07, + "loss": 0.0066, + "reward": 1.3737208843231201, + "reward_std": 0.054402053356170654, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3737209439277649, + "rewards/pad": 0.0, + "step": 1364 + }, + { + "completion_length": 45.265625, + "epoch": 0.4349904397705545, + "grad_norm": 22.74249267578125, + "kl": 0.205078125, + "learning_rate": 5.650095602294455e-07, + "loss": 0.0082, + "reward": 1.6794965267181396, + "reward_std": 0.0682016983628273, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5544965267181396, + "rewards/pad": 0.125, + "step": 1365 + }, + { + "completion_length": 97.828125, + "epoch": 0.43530911408540474, + "grad_norm": 39.36299514770508, + "kl": 0.1474609375, + "learning_rate": 5.646908859145952e-07, + "loss": 0.0059, + "reward": 1.3101704120635986, + "reward_std": 0.04966102913022041, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.31017038226127625, + "rewards/pad": 0.0, + "step": 1366 + }, + { + "completion_length": 74.1875, + "epoch": 0.43562778840025496, + "grad_norm": 39.882850646972656, + "kl": 0.12890625, + "learning_rate": 5.64372211599745e-07, + "loss": 0.0052, + "reward": 1.8632087707519531, + "reward_std": 0.07874581217765808, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6132087111473083, + "rewards/pad": 0.25, + "step": 1367 + }, + { + "completion_length": 97.734375, + "epoch": 0.4359464627151052, + "grad_norm": 41.4481086730957, + "kl": 0.10986328125, + "learning_rate": 5.640535372848948e-07, + "loss": 0.0044, + "reward": 1.6479605436325073, + "reward_std": 0.108167365193367, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6479606032371521, + "rewards/pad": 0.0, + "step": 1368 + }, + { + "completion_length": 124.09375, + "epoch": 0.4362651370299554, + "grad_norm": 43.37957763671875, + "kl": 0.2275390625, + "learning_rate": 5.637348629700446e-07, + "loss": 0.0091, + "reward": 1.5912022590637207, + "reward_std": 0.07037024199962616, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4662023186683655, + "step": 1369 + }, + { + "completion_length": 97.671875, + "epoch": 0.4365838113448056, + "grad_norm": 23.112123489379883, + "kl": 0.224609375, + "learning_rate": 5.634161886551943e-07, + "loss": 0.009, + "reward": 1.5291333198547363, + "reward_std": 0.0721755251288414, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5291333794593811, + "rewards/pad": 0.0, + "step": 1370 + }, + { + "completion_length": 72.109375, + "epoch": 0.43690248565965584, + "grad_norm": 44.03460693359375, + "kl": 0.10595703125, + "learning_rate": 5.630975143403441e-07, + "loss": 0.0042, + "reward": 1.6287412643432617, + "reward_std": 0.04039338603615761, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3787413537502289, + "step": 1371 + }, + { + "completion_length": 149.09375, + "epoch": 0.43722115997450606, + "grad_norm": 11.578285217285156, + "kl": 0.10302734375, + "learning_rate": 5.627788400254939e-07, + "loss": 0.0041, + "reward": 1.7141399383544922, + "reward_std": 0.05874771997332573, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.589139997959137, + "step": 1372 + }, + { + "completion_length": 20.15625, + "epoch": 0.4375398342893563, + "grad_norm": 55.78146743774414, + "kl": 0.330078125, + "learning_rate": 5.624601657106437e-07, + "loss": 0.0132, + "reward": 1.6458699703216553, + "reward_std": 0.1880033016204834, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.41149500012397766, + "rewards/pad": 0.25, + "step": 1373 + }, + { + "completion_length": 44.75, + "epoch": 0.4378585086042065, + "grad_norm": 41.31639099121094, + "kl": 0.2255859375, + "learning_rate": 5.621414913957934e-07, + "loss": 0.009, + "reward": 1.5588467121124268, + "reward_std": 0.08944406360387802, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.558846652507782, + "rewards/pad": 0.0, + "step": 1374 + }, + { + "completion_length": 121.328125, + "epoch": 0.4381771829190567, + "grad_norm": 23.76357650756836, + "kl": 0.1474609375, + "learning_rate": 5.618228170809432e-07, + "loss": 0.0059, + "reward": 1.5019047260284424, + "reward_std": 0.06296569854021072, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5019047260284424, + "step": 1375 + }, + { + "completion_length": 74.5625, + "epoch": 0.43849585723390694, + "grad_norm": 27.09462547302246, + "kl": 0.0947265625, + "learning_rate": 5.615041427660931e-07, + "loss": 0.0038, + "reward": 1.8667649030685425, + "reward_std": 0.054511696100234985, + "rewards/answer_reward": 0.375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4917648136615753, + "step": 1376 + }, + { + "completion_length": 98.59375, + "epoch": 0.43881453154875716, + "grad_norm": 49.495826721191406, + "kl": 0.2470703125, + "learning_rate": 5.611854684512429e-07, + "loss": 0.0099, + "reward": 1.688317060470581, + "reward_std": 0.06171800568699837, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4383169114589691, + "step": 1377 + }, + { + "completion_length": 22.96875, + "epoch": 0.4391332058636074, + "grad_norm": 264.8526306152344, + "kl": 0.158203125, + "learning_rate": 5.608667941363926e-07, + "loss": 0.0063, + "reward": 1.7475816011428833, + "reward_std": 0.17028731107711792, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.4819566011428833, + "rewards/pad": 0.28125, + "step": 1378 + }, + { + "completion_length": 176.09375, + "epoch": 0.4394518801784576, + "grad_norm": 17.579050064086914, + "kl": 0.056640625, + "learning_rate": 5.605481198215424e-07, + "loss": 0.0023, + "reward": 1.4881428480148315, + "reward_std": 0.0604608952999115, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.36314287781715393, + "step": 1379 + }, + { + "completion_length": 69.96875, + "epoch": 0.4397705544933078, + "grad_norm": 35.04433822631836, + "kl": 0.0947265625, + "learning_rate": 5.602294455066922e-07, + "loss": 0.0038, + "reward": 1.7537932395935059, + "reward_std": 0.10242816805839539, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6287932991981506, + "rewards/pad": 0.125, + "step": 1380 + }, + { + "completion_length": 70.015625, + "epoch": 0.44008922880815804, + "grad_norm": 39.497703552246094, + "kl": 0.12109375, + "learning_rate": 5.59910771191842e-07, + "loss": 0.0048, + "reward": 1.5401318073272705, + "reward_std": 0.0786762535572052, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.41513165831565857, + "step": 1381 + }, + { + "completion_length": 120.921875, + "epoch": 0.44040790312300826, + "grad_norm": 26.43705940246582, + "kl": 0.1396484375, + "learning_rate": 5.595920968769917e-07, + "loss": 0.0056, + "reward": 1.4561233520507812, + "reward_std": 0.054390233010053635, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.45612335205078125, + "rewards/pad": 0.0, + "step": 1382 + }, + { + "completion_length": 124.78125, + "epoch": 0.44072657743785854, + "grad_norm": 18.053184509277344, + "kl": 0.18359375, + "learning_rate": 5.592734225621415e-07, + "loss": 0.0073, + "reward": 1.66341233253479, + "reward_std": 0.0625927671790123, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5384123921394348, + "step": 1383 + }, + { + "completion_length": 70.015625, + "epoch": 0.44104525175270876, + "grad_norm": 1104.785888671875, + "kl": 0.138671875, + "learning_rate": 5.589547482472913e-07, + "loss": 0.0056, + "reward": 1.4395288228988647, + "reward_std": 0.048976339399814606, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.43952885270118713, + "step": 1384 + }, + { + "completion_length": 146.375, + "epoch": 0.441363926067559, + "grad_norm": 79.37278747558594, + "kl": 0.0869140625, + "learning_rate": 5.586360739324411e-07, + "loss": 0.0035, + "reward": 1.4432659149169922, + "reward_std": 0.04890685901045799, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.44326597452163696, + "step": 1385 + }, + { + "completion_length": 43.015625, + "epoch": 0.4416826003824092, + "grad_norm": 27.281627655029297, + "kl": 0.1533203125, + "learning_rate": 5.583173996175908e-07, + "loss": 0.0061, + "reward": 1.5859811305999756, + "reward_std": 0.07883980870246887, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5859811305999756, + "step": 1386 + }, + { + "completion_length": 44.65625, + "epoch": 0.4420012746972594, + "grad_norm": 37.530025482177734, + "kl": 0.1708984375, + "learning_rate": 5.579987253027406e-07, + "loss": 0.0068, + "reward": 1.510833740234375, + "reward_std": 0.09606379270553589, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.38583362102508545, + "rewards/pad": 0.125, + "step": 1387 + }, + { + "completion_length": 145.34375, + "epoch": 0.44231994901210964, + "grad_norm": 12.873120307922363, + "kl": 0.07275390625, + "learning_rate": 5.576800509878904e-07, + "loss": 0.0029, + "reward": 1.631828784942627, + "reward_std": 0.04249340295791626, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.631828784942627, + "rewards/pad": 0.0, + "step": 1388 + }, + { + "completion_length": 46.34375, + "epoch": 0.44263862332695986, + "grad_norm": 40.11998748779297, + "kl": 0.3203125, + "learning_rate": 5.573613766730401e-07, + "loss": 0.0129, + "reward": 1.6404682397842407, + "reward_std": 0.13934734463691711, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4373432695865631, + "rewards/pad": 0.203125, + "step": 1389 + }, + { + "completion_length": 97.78125, + "epoch": 0.4429572976418101, + "grad_norm": 41.30707550048828, + "kl": 0.099609375, + "learning_rate": 5.570427023581899e-07, + "loss": 0.004, + "reward": 1.713468313217163, + "reward_std": 0.13982240855693817, + "rewards/answer_reward": 0.328125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.3853432238101959, + "step": 1390 + }, + { + "completion_length": 20.421875, + "epoch": 0.4432759719566603, + "grad_norm": 64.35762023925781, + "kl": 0.30859375, + "learning_rate": 5.567240280433397e-07, + "loss": 0.0124, + "reward": 1.66224205493927, + "reward_std": 0.15928740799427032, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.61536705493927, + "rewards/pad": 0.046875, + "step": 1391 + }, + { + "completion_length": 67.71875, + "epoch": 0.4435946462715105, + "grad_norm": 18.37198257446289, + "kl": 0.1513671875, + "learning_rate": 5.564053537284895e-07, + "loss": 0.006, + "reward": 1.6161123514175415, + "reward_std": 0.1390165537595749, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5379872918128967, + "rewards/pad": 0.078125, + "step": 1392 + }, + { + "completion_length": 69.65625, + "epoch": 0.44391332058636074, + "grad_norm": 97.37084197998047, + "kl": 0.1484375, + "learning_rate": 5.560866794136391e-07, + "loss": 0.0059, + "reward": 1.722306728363037, + "reward_std": 0.13341601192951202, + "rewards/answer_reward": 0.171875, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5504317283630371, + "step": 1393 + }, + { + "completion_length": 122.078125, + "epoch": 0.44423199490121096, + "grad_norm": 12.7796630859375, + "kl": 0.08642578125, + "learning_rate": 5.557680050987889e-07, + "loss": 0.0035, + "reward": 1.5103886127471924, + "reward_std": 0.0922664999961853, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.5260135531425476, + "step": 1394 + }, + { + "completion_length": 121.90625, + "epoch": 0.4445506692160612, + "grad_norm": 13.822649002075195, + "kl": 0.11279296875, + "learning_rate": 5.554493307839388e-07, + "loss": 0.0045, + "reward": 1.471165657043457, + "reward_std": 0.0489761158823967, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.47116556763648987, + "step": 1395 + }, + { + "completion_length": 122.265625, + "epoch": 0.4448693435309114, + "grad_norm": 95.01806640625, + "kl": 0.1142578125, + "learning_rate": 5.551306564690886e-07, + "loss": 0.0046, + "reward": 1.5615136623382568, + "reward_std": 0.04147946089506149, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5615136027336121, + "rewards/pad": 0.0, + "step": 1396 + }, + { + "completion_length": 70.921875, + "epoch": 0.4451880178457616, + "grad_norm": 40.63417053222656, + "kl": 0.142578125, + "learning_rate": 5.548119821542383e-07, + "loss": 0.0057, + "reward": 1.5273734331130981, + "reward_std": 0.040808022022247314, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5273733735084534, + "rewards/pad": 0.0, + "step": 1397 + }, + { + "completion_length": 70.640625, + "epoch": 0.44550669216061184, + "grad_norm": 47.04145812988281, + "kl": 0.169921875, + "learning_rate": 5.544933078393881e-07, + "loss": 0.0068, + "reward": 1.4421722888946533, + "reward_std": 0.056907620280981064, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.44217219948768616, + "step": 1398 + }, + { + "completion_length": 72.125, + "epoch": 0.44582536647546206, + "grad_norm": 104.2425308227539, + "kl": 0.15234375, + "learning_rate": 5.541746335245379e-07, + "loss": 0.0061, + "reward": 1.5128662586212158, + "reward_std": 0.09221340715885162, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4972413182258606, + "rewards/pad": 0.015625, + "step": 1399 + }, + { + "completion_length": 46.1875, + "epoch": 0.4461440407903123, + "grad_norm": 37.326927185058594, + "kl": 0.2197265625, + "learning_rate": 5.538559592096877e-07, + "loss": 0.0088, + "reward": 1.8115777969360352, + "reward_std": 0.052388474345207214, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5615776777267456, + "step": 1400 + }, + { + "completion_length": 70.453125, + "epoch": 0.4464627151051625, + "grad_norm": 129.155517578125, + "kl": 0.1787109375, + "learning_rate": 5.535372848948374e-07, + "loss": 0.0072, + "reward": 1.4227665662765503, + "reward_std": 0.0634697824716568, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4227665066719055, + "step": 1401 + }, + { + "completion_length": 122.609375, + "epoch": 0.4467813894200127, + "grad_norm": 6.261690139770508, + "kl": 0.14453125, + "learning_rate": 5.532186105799872e-07, + "loss": 0.0058, + "reward": 1.3813254833221436, + "reward_std": 0.06035871058702469, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3813254237174988, + "rewards/pad": 0.0, + "step": 1402 + }, + { + "completion_length": 98.59375, + "epoch": 0.447100063734863, + "grad_norm": 69.14564514160156, + "kl": 0.1015625, + "learning_rate": 5.52899936265137e-07, + "loss": 0.0041, + "reward": 1.9547457695007324, + "reward_std": 0.12339489161968231, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5797458291053772, + "rewards/pad": 0.375, + "step": 1403 + }, + { + "completion_length": 48.609375, + "epoch": 0.4474187380497132, + "grad_norm": 79.41541290283203, + "kl": 0.16015625, + "learning_rate": 5.525812619502868e-07, + "loss": 0.0064, + "reward": 1.612067461013794, + "reward_std": 0.11976909637451172, + "rewards/pad": 0.234375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.37769243121147156, + "step": 1404 + }, + { + "completion_length": 70.75, + "epoch": 0.44773741236456344, + "grad_norm": 24.919973373413086, + "kl": 0.1552734375, + "learning_rate": 5.522625876354365e-07, + "loss": 0.0062, + "reward": 1.6123287677764893, + "reward_std": 0.07202111184597015, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.48732879757881165, + "step": 1405 + }, + { + "completion_length": 71.109375, + "epoch": 0.44805608667941366, + "grad_norm": 53.67876052856445, + "kl": 0.1396484375, + "learning_rate": 5.519439133205863e-07, + "loss": 0.0056, + "reward": 1.5918638706207275, + "reward_std": 0.07275165617465973, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.46686387062072754, + "step": 1406 + }, + { + "completion_length": 95.375, + "epoch": 0.4483747609942639, + "grad_norm": 20.173824310302734, + "kl": 0.369140625, + "learning_rate": 5.516252390057361e-07, + "loss": 0.0148, + "reward": 1.5206084251403809, + "reward_std": 0.13873018324375153, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.5362333655357361, + "rewards/pad": 0.0, + "step": 1407 + }, + { + "completion_length": 120.9375, + "epoch": 0.4486934353091141, + "grad_norm": 25.047237396240234, + "kl": 0.10107421875, + "learning_rate": 5.513065646908859e-07, + "loss": 0.004, + "reward": 1.4869604110717773, + "reward_std": 0.061503179371356964, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3619605302810669, + "step": 1408 + }, + { + "completion_length": 46.96875, + "epoch": 0.4490121096239643, + "grad_norm": 51.29338455200195, + "kl": 0.146484375, + "learning_rate": 5.509878903760356e-07, + "loss": 0.0059, + "reward": 1.7304821014404297, + "reward_std": 0.12294122576713562, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5586071014404297, + "rewards/pad": 0.171875, + "step": 1409 + }, + { + "completion_length": 46.390625, + "epoch": 0.44933078393881454, + "grad_norm": 25.94217300415039, + "kl": 0.1142578125, + "learning_rate": 5.506692160611854e-07, + "loss": 0.0046, + "reward": 1.7969729900360107, + "reward_std": 0.12597975134849548, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.43759799003601074, + "rewards/pad": 0.359375, + "step": 1410 + }, + { + "completion_length": 95.90625, + "epoch": 0.44964945825366476, + "grad_norm": 17.947458267211914, + "kl": 0.2578125, + "learning_rate": 5.503505417463352e-07, + "loss": 0.0103, + "reward": 1.6846892833709717, + "reward_std": 0.07526648789644241, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5596892237663269, + "step": 1411 + }, + { + "completion_length": 146.140625, + "epoch": 0.449968132568515, + "grad_norm": 42.996315002441406, + "kl": 0.130859375, + "learning_rate": 5.50031867431485e-07, + "loss": 0.0053, + "reward": 1.3987228870391846, + "reward_std": 0.10691266506910324, + "rewards/pad": 0.015625, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.38309788703918457, + "step": 1412 + }, + { + "completion_length": 71.703125, + "epoch": 0.4502868068833652, + "grad_norm": 42.76258850097656, + "kl": 0.2138671875, + "learning_rate": 5.497131931166347e-07, + "loss": 0.0085, + "reward": 1.5707018375396729, + "reward_std": 0.07505904138088226, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.44570183753967285, + "step": 1413 + }, + { + "completion_length": 72.859375, + "epoch": 0.4506054811982154, + "grad_norm": 47.79791259765625, + "kl": 0.1513671875, + "learning_rate": 5.493945188017846e-07, + "loss": 0.0061, + "reward": 1.6459414958953857, + "reward_std": 0.08020886778831482, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5209414958953857, + "step": 1414 + }, + { + "completion_length": 172.125, + "epoch": 0.45092415551306564, + "grad_norm": 71.31310272216797, + "kl": 0.0751953125, + "learning_rate": 5.490758444869344e-07, + "loss": 0.003, + "reward": 1.658172845840454, + "reward_std": 0.06371012330055237, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5331727862358093, + "rewards/pad": 0.125, + "step": 1415 + }, + { + "completion_length": 150.515625, + "epoch": 0.45124282982791586, + "grad_norm": 8.224554061889648, + "kl": 0.1787109375, + "learning_rate": 5.487571701720841e-07, + "loss": 0.0071, + "reward": 1.5972743034362793, + "reward_std": 0.1921130120754242, + "rewards/pad": 0.109375, + "rewards/tracking_format_reward": 0.96875, + "rewards/tracking_iou_reward": 0.5191493630409241, + "step": 1416 + }, + { + "completion_length": 149.9375, + "epoch": 0.4515615041427661, + "grad_norm": 12.27308464050293, + "kl": 0.0810546875, + "learning_rate": 5.484384958572339e-07, + "loss": 0.0032, + "reward": 1.4329180717468262, + "reward_std": 0.047323573380708694, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4329180121421814, + "step": 1417 + }, + { + "completion_length": 120.609375, + "epoch": 0.4518801784576163, + "grad_norm": 14.535323143005371, + "kl": 0.177734375, + "learning_rate": 5.481198215423837e-07, + "loss": 0.0071, + "reward": 1.3263839483261108, + "reward_std": 0.029305964708328247, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3263840079307556, + "rewards/pad": 0.0, + "step": 1418 + }, + { + "completion_length": 123.9375, + "epoch": 0.4521988527724665, + "grad_norm": 74.81829833984375, + "kl": 0.10595703125, + "learning_rate": 5.478011472275335e-07, + "loss": 0.0042, + "reward": 1.645141363143921, + "reward_std": 0.09291210770606995, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5201413035392761, + "rewards/pad": 0.125, + "step": 1419 + }, + { + "completion_length": 94.421875, + "epoch": 0.45251752708731674, + "grad_norm": 66.59827423095703, + "kl": 0.099609375, + "learning_rate": 5.474824729126832e-07, + "loss": 0.004, + "reward": 1.3246803283691406, + "reward_std": 0.11868653446435928, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.3403053283691406, + "rewards/pad": 0.0, + "step": 1420 + }, + { + "completion_length": 121.921875, + "epoch": 0.45283620140216696, + "grad_norm": 23.449520111083984, + "kl": 0.1650390625, + "learning_rate": 5.47163798597833e-07, + "loss": 0.0066, + "reward": 1.550499439239502, + "reward_std": 0.1245659664273262, + "rewards/pad": 0.109375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.44112440943717957, + "step": 1421 + }, + { + "completion_length": 44.390625, + "epoch": 0.45315487571701724, + "grad_norm": 54.22697830200195, + "kl": 0.44921875, + "learning_rate": 5.468451242829828e-07, + "loss": 0.018, + "reward": 1.5884101390838623, + "reward_std": 0.1501428335905075, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5884101986885071, + "rewards/pad": 0.0, + "step": 1422 + }, + { + "completion_length": 97.984375, + "epoch": 0.45347355003186746, + "grad_norm": 19.764827728271484, + "kl": 0.1181640625, + "learning_rate": 5.465264499681326e-07, + "loss": 0.0047, + "reward": 1.6852483749389648, + "reward_std": 0.07374419271945953, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5602483749389648, + "step": 1423 + }, + { + "completion_length": 72.671875, + "epoch": 0.4537922243467177, + "grad_norm": 97.86742401123047, + "kl": 0.150390625, + "learning_rate": 5.462077756532823e-07, + "loss": 0.006, + "reward": 1.7004663944244385, + "reward_std": 0.17683261632919312, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5598414540290833, + "rewards/pad": 0.140625, + "step": 1424 + }, + { + "completion_length": 172.28125, + "epoch": 0.4541108986615679, + "grad_norm": 6.176180839538574, + "kl": 0.12109375, + "learning_rate": 5.458891013384321e-07, + "loss": 0.0048, + "reward": 1.4284148216247559, + "reward_std": 0.10202738642692566, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.444039911031723, + "step": 1425 + }, + { + "completion_length": 97.546875, + "epoch": 0.4544295729764181, + "grad_norm": 84.13567352294922, + "kl": 0.1826171875, + "learning_rate": 5.455704270235819e-07, + "loss": 0.0073, + "reward": 1.5227749347686768, + "reward_std": 0.07707509398460388, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.39777496457099915, + "step": 1426 + }, + { + "completion_length": 71.953125, + "epoch": 0.45474824729126834, + "grad_norm": 48.81588363647461, + "kl": 0.2041015625, + "learning_rate": 5.452517527087317e-07, + "loss": 0.0082, + "reward": 1.6935824155807495, + "reward_std": 0.09216948598623276, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5685823559761047, + "step": 1427 + }, + { + "completion_length": 95.390625, + "epoch": 0.45506692160611856, + "grad_norm": 26.522541046142578, + "kl": 0.1357421875, + "learning_rate": 5.449330783938814e-07, + "loss": 0.0054, + "reward": 1.5668354034423828, + "reward_std": 0.07392378896474838, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5668354034423828, + "rewards/pad": 0.0, + "step": 1428 + }, + { + "completion_length": 148.890625, + "epoch": 0.4553855959209688, + "grad_norm": 11.638447761535645, + "kl": 0.09228515625, + "learning_rate": 5.446144040790312e-07, + "loss": 0.0037, + "reward": 1.5445245504379272, + "reward_std": 0.037910446524620056, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.419524610042572, + "rewards/pad": 0.125, + "step": 1429 + }, + { + "completion_length": 71.625, + "epoch": 0.455704270235819, + "grad_norm": 57.1502685546875, + "kl": 0.1708984375, + "learning_rate": 5.44295729764181e-07, + "loss": 0.0068, + "reward": 1.686220407485962, + "reward_std": 0.07011242210865021, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5612203478813171, + "rewards/pad": 0.125, + "step": 1430 + }, + { + "completion_length": 123.578125, + "epoch": 0.4560229445506692, + "grad_norm": 33.73246383666992, + "kl": 0.123046875, + "learning_rate": 5.439770554493309e-07, + "loss": 0.0049, + "reward": 1.6060597896575928, + "reward_std": 0.058518167585134506, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4810597896575928, + "step": 1431 + }, + { + "completion_length": 128.859375, + "epoch": 0.45634161886551944, + "grad_norm": 49.34904479980469, + "kl": 0.076171875, + "learning_rate": 5.436583811344804e-07, + "loss": 0.003, + "reward": 1.4976006746292114, + "reward_std": 0.06735117733478546, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.37260061502456665, + "rewards/pad": 0.125, + "step": 1432 + }, + { + "completion_length": 99.328125, + "epoch": 0.45666029318036966, + "grad_norm": 19.230791091918945, + "kl": 0.09375, + "learning_rate": 5.433397068196303e-07, + "loss": 0.0037, + "reward": 1.8719754219055176, + "reward_std": 0.08034700155258179, + "rewards/answer_reward": 0.484375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.3876003623008728, + "step": 1433 + }, + { + "completion_length": 121.53125, + "epoch": 0.4569789674952199, + "grad_norm": 14.015466690063477, + "kl": 0.1376953125, + "learning_rate": 5.430210325047801e-07, + "loss": 0.0055, + "reward": 1.534825086593628, + "reward_std": 0.04365585371851921, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5348250865936279, + "step": 1434 + }, + { + "completion_length": 70.84375, + "epoch": 0.4572976418100701, + "grad_norm": 18.4355525970459, + "kl": 0.1376953125, + "learning_rate": 5.427023581899299e-07, + "loss": 0.0055, + "reward": 1.6791949272155762, + "reward_std": 0.04616924747824669, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5541948676109314, + "step": 1435 + }, + { + "completion_length": 93.65625, + "epoch": 0.4576163161249203, + "grad_norm": 16.46039581298828, + "kl": 0.26953125, + "learning_rate": 5.423836838750796e-07, + "loss": 0.0108, + "reward": 1.4510504007339478, + "reward_std": 0.0750553086400032, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.45105046033859253, + "step": 1436 + }, + { + "completion_length": 97.390625, + "epoch": 0.45793499043977054, + "grad_norm": 23.34442138671875, + "kl": 0.1474609375, + "learning_rate": 5.420650095602294e-07, + "loss": 0.0059, + "reward": 1.5739753246307373, + "reward_std": 0.09569709002971649, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5739753246307373, + "rewards/pad": 0.0, + "step": 1437 + }, + { + "completion_length": 147.15625, + "epoch": 0.45825366475462076, + "grad_norm": 27.796993255615234, + "kl": 0.091796875, + "learning_rate": 5.417463352453792e-07, + "loss": 0.0037, + "reward": 1.5219330787658691, + "reward_std": 0.024136081337928772, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5219330787658691, + "rewards/pad": 0.0, + "step": 1438 + }, + { + "completion_length": 71.125, + "epoch": 0.458572339069471, + "grad_norm": 45.78583526611328, + "kl": 0.119140625, + "learning_rate": 5.41427660930529e-07, + "loss": 0.0048, + "reward": 1.8081820011138916, + "reward_std": 0.08273088932037354, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5581819415092468, + "rewards/pad": 0.25, + "step": 1439 + }, + { + "completion_length": 125.765625, + "epoch": 0.4588910133843212, + "grad_norm": 23.17354965209961, + "kl": 0.1083984375, + "learning_rate": 5.411089866156787e-07, + "loss": 0.0043, + "reward": 1.6552670001983643, + "reward_std": 0.20718078315258026, + "rewards/pad": 0.234375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.42089203000068665, + "step": 1440 + }, + { + "completion_length": 123.9375, + "epoch": 0.4592096876991714, + "grad_norm": 23.839529037475586, + "kl": 0.25, + "learning_rate": 5.407903123008285e-07, + "loss": 0.01, + "reward": 1.5547746419906616, + "reward_std": 0.07701779901981354, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4297747313976288, + "rewards/pad": 0.125, + "step": 1441 + }, + { + "completion_length": 45.78125, + "epoch": 0.4595283620140217, + "grad_norm": 40.79808807373047, + "kl": 0.2197265625, + "learning_rate": 5.404716379859783e-07, + "loss": 0.0088, + "reward": 1.6783041954040527, + "reward_std": 0.12530480325222015, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5533040761947632, + "step": 1442 + }, + { + "completion_length": 74.5625, + "epoch": 0.4598470363288719, + "grad_norm": 43.00371170043945, + "kl": 0.1533203125, + "learning_rate": 5.401529636711281e-07, + "loss": 0.0061, + "reward": 1.6545014381408691, + "reward_std": 0.11320410668849945, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.38887643814086914, + "rewards/pad": 0.265625, + "step": 1443 + }, + { + "completion_length": 99.296875, + "epoch": 0.46016571064372214, + "grad_norm": 16.58584976196289, + "kl": 0.1318359375, + "learning_rate": 5.398342893562778e-07, + "loss": 0.0053, + "reward": 1.5221130847930908, + "reward_std": 0.08949887752532959, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5221132636070251, + "rewards/pad": 0.0, + "step": 1444 + }, + { + "completion_length": 46.34375, + "epoch": 0.46048438495857236, + "grad_norm": 67.59211730957031, + "kl": 0.1787109375, + "learning_rate": 5.395156150414276e-07, + "loss": 0.0072, + "reward": 1.4442793130874634, + "reward_std": 0.1727423369884491, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3661543130874634, + "rewards/pad": 0.078125, + "step": 1445 + }, + { + "completion_length": 124.34375, + "epoch": 0.4608030592734226, + "grad_norm": 34.86455535888672, + "kl": 0.1201171875, + "learning_rate": 5.391969407265774e-07, + "loss": 0.0048, + "reward": 1.561646819114685, + "reward_std": 0.07751183211803436, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5460219383239746, + "rewards/pad": 0.015625, + "step": 1446 + }, + { + "completion_length": 125.765625, + "epoch": 0.4611217335882728, + "grad_norm": 31.93128776550293, + "kl": 0.1328125, + "learning_rate": 5.388782664117271e-07, + "loss": 0.0053, + "reward": 1.6424591541290283, + "reward_std": 0.1986120492219925, + "rewards/pad": 0.21875, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.43933412432670593, + "step": 1447 + }, + { + "completion_length": 148.828125, + "epoch": 0.461440407903123, + "grad_norm": 21.80022430419922, + "kl": 0.0849609375, + "learning_rate": 5.385595920968769e-07, + "loss": 0.0034, + "reward": 1.6194400787353516, + "reward_std": 0.09938361495733261, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5256900787353516, + "rewards/pad": 0.09375, + "step": 1448 + }, + { + "completion_length": 96.953125, + "epoch": 0.46175908221797324, + "grad_norm": 40.039730072021484, + "kl": 0.15625, + "learning_rate": 5.382409177820267e-07, + "loss": 0.0063, + "reward": 1.564121961593628, + "reward_std": 0.11281084269285202, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5484969615936279, + "rewards/pad": 0.015625, + "step": 1449 + }, + { + "completion_length": 127.578125, + "epoch": 0.46207775653282346, + "grad_norm": 19.242937088012695, + "kl": 0.10302734375, + "learning_rate": 5.379222434671765e-07, + "loss": 0.0041, + "reward": 1.6729660034179688, + "reward_std": 0.06923995912075043, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5479658246040344, + "rewards/pad": 0.125, + "step": 1450 + }, + { + "completion_length": 70.4375, + "epoch": 0.4623964308476737, + "grad_norm": 26.17292022705078, + "kl": 0.1796875, + "learning_rate": 5.376035691523262e-07, + "loss": 0.0072, + "reward": 1.6035188436508179, + "reward_std": 0.124393530189991, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.49414384365081787, + "rewards/pad": 0.109375, + "step": 1451 + }, + { + "completion_length": 44.4375, + "epoch": 0.4627151051625239, + "grad_norm": 101.223388671875, + "kl": 0.2099609375, + "learning_rate": 5.372848948374761e-07, + "loss": 0.0084, + "reward": 1.464913010597229, + "reward_std": 0.07390671968460083, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.46491295099258423, + "rewards/pad": 0.0, + "step": 1452 + }, + { + "completion_length": 123.234375, + "epoch": 0.4630337794773741, + "grad_norm": 25.080516815185547, + "kl": 0.1220703125, + "learning_rate": 5.369662205226259e-07, + "loss": 0.0049, + "reward": 1.477994680404663, + "reward_std": 0.0637764185667038, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3529945909976959, + "step": 1453 + }, + { + "completion_length": 122.359375, + "epoch": 0.46335245379222434, + "grad_norm": 20.0180606842041, + "kl": 0.1083984375, + "learning_rate": 5.366475462077757e-07, + "loss": 0.0043, + "reward": 1.2684621810913086, + "reward_std": 0.048859499394893646, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.26846227049827576, + "rewards/pad": 0.0, + "step": 1454 + }, + { + "completion_length": 71.84375, + "epoch": 0.46367112810707456, + "grad_norm": 15.79022216796875, + "kl": 0.2197265625, + "learning_rate": 5.363288718929254e-07, + "loss": 0.0088, + "reward": 1.782228708267212, + "reward_std": 0.1331326961517334, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6572288274765015, + "rewards/pad": 0.125, + "step": 1455 + }, + { + "completion_length": 70.78125, + "epoch": 0.4639898024219248, + "grad_norm": 242.91552734375, + "kl": 0.1376953125, + "learning_rate": 5.360101975780752e-07, + "loss": 0.0055, + "reward": 1.8918707370758057, + "reward_std": 0.061253033578395844, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.6418708562850952, + "step": 1456 + }, + { + "completion_length": 73.171875, + "epoch": 0.464308476736775, + "grad_norm": 13.989324569702148, + "kl": 0.1298828125, + "learning_rate": 5.35691523263225e-07, + "loss": 0.0052, + "reward": 1.6837668418884277, + "reward_std": 0.061112187802791595, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.43376684188842773, + "step": 1457 + }, + { + "completion_length": 126.46875, + "epoch": 0.4646271510516252, + "grad_norm": 31.25027847290039, + "kl": 0.099609375, + "learning_rate": 5.353728489483748e-07, + "loss": 0.004, + "reward": 1.5163158178329468, + "reward_std": 0.06596622616052628, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.2663158178329468, + "step": 1458 + }, + { + "completion_length": 70.953125, + "epoch": 0.46494582536647544, + "grad_norm": 24.366724014282227, + "kl": 0.2314453125, + "learning_rate": 5.350541746335245e-07, + "loss": 0.0092, + "reward": 1.5501924753189087, + "reward_std": 0.08063836395740509, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5501924753189087, + "rewards/pad": 0.0, + "step": 1459 + }, + { + "completion_length": 126.8125, + "epoch": 0.46526449968132566, + "grad_norm": 18.02052116394043, + "kl": 0.1611328125, + "learning_rate": 5.347355003186743e-07, + "loss": 0.0065, + "reward": 1.519972801208496, + "reward_std": 0.06053301319479942, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5199728012084961, + "step": 1460 + }, + { + "completion_length": 124.203125, + "epoch": 0.4655831739961759, + "grad_norm": 11.933880805969238, + "kl": 0.10205078125, + "learning_rate": 5.344168260038241e-07, + "loss": 0.0041, + "reward": 1.4709376096725464, + "reward_std": 0.10856231302022934, + "rewards/answer_reward": 0.046875, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.42406266927719116, + "step": 1461 + }, + { + "completion_length": 124.484375, + "epoch": 0.46590184831102616, + "grad_norm": 9.68100643157959, + "kl": 0.1279296875, + "learning_rate": 5.340981516889739e-07, + "loss": 0.0051, + "reward": 1.5821932554244995, + "reward_std": 0.08199529349803925, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4571932554244995, + "rewards/pad": 0.125, + "step": 1462 + }, + { + "completion_length": 48.171875, + "epoch": 0.4662205226258764, + "grad_norm": 30.872400283813477, + "kl": 0.2138671875, + "learning_rate": 5.337794773741236e-07, + "loss": 0.0085, + "reward": 1.7756941318511963, + "reward_std": 0.06104860454797745, + "rewards/answer_reward": 0.375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4006942808628082, + "step": 1463 + }, + { + "completion_length": 46.140625, + "epoch": 0.4665391969407266, + "grad_norm": 32.10598373413086, + "kl": 0.1533203125, + "learning_rate": 5.334608030592734e-07, + "loss": 0.0061, + "reward": 1.8779191970825195, + "reward_std": 0.09105006605386734, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6279191374778748, + "rewards/pad": 0.25, + "step": 1464 + }, + { + "completion_length": 95.3125, + "epoch": 0.4668578712555768, + "grad_norm": 31.138216018676758, + "kl": 0.2177734375, + "learning_rate": 5.331421287444232e-07, + "loss": 0.0087, + "reward": 1.5052273273468018, + "reward_std": 0.05461052432656288, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5052273869514465, + "rewards/pad": 0.0, + "step": 1465 + }, + { + "completion_length": 69.796875, + "epoch": 0.46717654557042704, + "grad_norm": 29.701608657836914, + "kl": 0.15625, + "learning_rate": 5.32823454429573e-07, + "loss": 0.0062, + "reward": 1.4903333187103271, + "reward_std": 0.07421676814556122, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.49033334851264954, + "step": 1466 + }, + { + "completion_length": 71.09375, + "epoch": 0.46749521988527726, + "grad_norm": 21.280033111572266, + "kl": 0.1953125, + "learning_rate": 5.325047801147227e-07, + "loss": 0.0078, + "reward": 1.7816627025604248, + "reward_std": 0.06757558137178421, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.6566627621650696, + "step": 1467 + }, + { + "completion_length": 150.796875, + "epoch": 0.4678138942001275, + "grad_norm": 10.08774185180664, + "kl": 0.12255859375, + "learning_rate": 5.321861057998725e-07, + "loss": 0.0049, + "reward": 1.463547945022583, + "reward_std": 0.03903178498148918, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.33854788541793823, + "rewards/pad": 0.125, + "step": 1468 + }, + { + "completion_length": 71.421875, + "epoch": 0.4681325685149777, + "grad_norm": 31.640708923339844, + "kl": 0.1484375, + "learning_rate": 5.318674314850224e-07, + "loss": 0.0059, + "reward": 1.387979507446289, + "reward_std": 0.0799902081489563, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.37235450744628906, + "rewards/pad": 0.015625, + "step": 1469 + }, + { + "completion_length": 126.125, + "epoch": 0.4684512428298279, + "grad_norm": 63.39480972290039, + "kl": 0.1064453125, + "learning_rate": 5.315487571701722e-07, + "loss": 0.0043, + "reward": 1.4995639324188232, + "reward_std": 0.14248573780059814, + "rewards/answer_reward": 0.0625, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.43706390261650085, + "step": 1470 + }, + { + "completion_length": 96.390625, + "epoch": 0.46876991714467814, + "grad_norm": 33.108638763427734, + "kl": 0.30859375, + "learning_rate": 5.312300828553218e-07, + "loss": 0.0124, + "reward": 1.610578179359436, + "reward_std": 0.11231406033039093, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.610578179359436, + "rewards/pad": 0.0, + "step": 1471 + }, + { + "completion_length": 70.453125, + "epoch": 0.46908859145952836, + "grad_norm": 17.15878677368164, + "kl": 0.201171875, + "learning_rate": 5.309114085404716e-07, + "loss": 0.008, + "reward": 1.852674126625061, + "reward_std": 0.06432235985994339, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.727674126625061, + "rewards/pad": 0.125, + "step": 1472 + }, + { + "completion_length": 71.703125, + "epoch": 0.4694072657743786, + "grad_norm": 29.73100471496582, + "kl": 0.158203125, + "learning_rate": 5.305927342256214e-07, + "loss": 0.0063, + "reward": 1.6606179475784302, + "reward_std": 0.08843652904033661, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5356178283691406, + "step": 1473 + }, + { + "completion_length": 149.21875, + "epoch": 0.4697259400892288, + "grad_norm": 8.382768630981445, + "kl": 0.11083984375, + "learning_rate": 5.302740599107712e-07, + "loss": 0.0044, + "reward": 1.3328176736831665, + "reward_std": 0.08705488592386246, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.3484426736831665, + "rewards/pad": 0.0, + "step": 1474 + }, + { + "completion_length": 72.109375, + "epoch": 0.470044614404079, + "grad_norm": 22.395475387573242, + "kl": 0.16015625, + "learning_rate": 5.299553855959209e-07, + "loss": 0.0064, + "reward": 1.524498701095581, + "reward_std": 0.044502221047878265, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.39949876070022583, + "rewards/pad": 0.125, + "step": 1475 + }, + { + "completion_length": 72.765625, + "epoch": 0.47036328871892924, + "grad_norm": 54.53713607788086, + "kl": 0.1708984375, + "learning_rate": 5.296367112810707e-07, + "loss": 0.0068, + "reward": 1.5576719045639038, + "reward_std": 0.23511239886283875, + "rewards/answer_reward": 0.21875, + "rewards/format_reward_gqa": 0.984375, + "rewards/iou_glue_reward": 0.3545469045639038, + "step": 1476 + }, + { + "completion_length": 97.15625, + "epoch": 0.47068196303377946, + "grad_norm": 13.809905052185059, + "kl": 0.2216796875, + "learning_rate": 5.293180369662205e-07, + "loss": 0.0089, + "reward": 1.5029902458190918, + "reward_std": 0.048063814640045166, + "rewards/answer_reward": 0.0, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5029902458190918, + "step": 1477 + }, + { + "completion_length": 147.171875, + "epoch": 0.4710006373486297, + "grad_norm": 30.669885635375977, + "kl": 0.08984375, + "learning_rate": 5.289993626513702e-07, + "loss": 0.0036, + "reward": 1.568947434425354, + "reward_std": 0.11746594309806824, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.459572434425354, + "step": 1478 + }, + { + "completion_length": 122.546875, + "epoch": 0.4713193116634799, + "grad_norm": 54.18355941772461, + "kl": 0.14453125, + "learning_rate": 5.2868068833652e-07, + "loss": 0.0058, + "reward": 1.4720231294631958, + "reward_std": 0.13342934846878052, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.4876481592655182, + "step": 1479 + }, + { + "completion_length": 72.015625, + "epoch": 0.4716379859783301, + "grad_norm": 16.47974395751953, + "kl": 0.181640625, + "learning_rate": 5.283620140216698e-07, + "loss": 0.0073, + "reward": 1.4542089700698853, + "reward_std": 0.08869212865829468, + "rewards/pad": 0.03125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.42295897006988525, + "step": 1480 + }, + { + "completion_length": 69.203125, + "epoch": 0.4719566602931804, + "grad_norm": 104.44915771484375, + "kl": 0.1279296875, + "learning_rate": 5.280433397068196e-07, + "loss": 0.0051, + "reward": 1.4982777833938599, + "reward_std": 0.06480859965085983, + "rewards/answer_reward": 0.0, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.49827784299850464, + "step": 1481 + }, + { + "completion_length": 74.0, + "epoch": 0.4722753346080306, + "grad_norm": 23.072843551635742, + "kl": 0.1982421875, + "learning_rate": 5.277246653919693e-07, + "loss": 0.0079, + "reward": 1.7503235340118408, + "reward_std": 0.06959592550992966, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6253235936164856, + "rewards/pad": 0.125, + "step": 1482 + }, + { + "completion_length": 123.09375, + "epoch": 0.47259400892288084, + "grad_norm": 13.988505363464355, + "kl": 0.111328125, + "learning_rate": 5.274059910771191e-07, + "loss": 0.0045, + "reward": 1.515568494796753, + "reward_std": 0.04018034413456917, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5155684351921082, + "rewards/pad": 0.0, + "step": 1483 + }, + { + "completion_length": 77.828125, + "epoch": 0.47291268323773106, + "grad_norm": 22.228260040283203, + "kl": 0.1337890625, + "learning_rate": 5.270873167622689e-07, + "loss": 0.0054, + "reward": 1.716814637184143, + "reward_std": 0.10153313726186752, + "rewards/answer_reward": 0.421875, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.29493963718414307, + "step": 1484 + }, + { + "completion_length": 72.21875, + "epoch": 0.4732313575525813, + "grad_norm": 66.78959655761719, + "kl": 0.134765625, + "learning_rate": 5.267686424474187e-07, + "loss": 0.0054, + "reward": 1.7643070220947266, + "reward_std": 0.1584181785583496, + "rewards/pad": 0.328125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.436181902885437, + "step": 1485 + }, + { + "completion_length": 99.515625, + "epoch": 0.4735500318674315, + "grad_norm": 15.219985008239746, + "kl": 0.205078125, + "learning_rate": 5.264499681325684e-07, + "loss": 0.0082, + "reward": 1.3323643207550049, + "reward_std": 0.07871396839618683, + "rewards/pad": 0.015625, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3167392909526825, + "step": 1486 + }, + { + "completion_length": 123.1875, + "epoch": 0.4738687061822817, + "grad_norm": 45.12118148803711, + "kl": 0.10400390625, + "learning_rate": 5.261312938177182e-07, + "loss": 0.0042, + "reward": 1.6009794473648071, + "reward_std": 0.13043075799942017, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.5853543877601624, + "rewards/pad": 0.03125, + "step": 1487 + }, + { + "completion_length": 122.671875, + "epoch": 0.47418738049713194, + "grad_norm": 18.35110092163086, + "kl": 0.103515625, + "learning_rate": 5.25812619502868e-07, + "loss": 0.0041, + "reward": 1.6021339893341064, + "reward_std": 0.0574474036693573, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4771340787410736, + "step": 1488 + }, + { + "completion_length": 70.421875, + "epoch": 0.47450605481198216, + "grad_norm": 30.877819061279297, + "kl": 0.15234375, + "learning_rate": 5.254939451880179e-07, + "loss": 0.0061, + "reward": 1.5548548698425293, + "reward_std": 0.08415327221155167, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4611048996448517, + "rewards/pad": 0.09375, + "step": 1489 + }, + { + "completion_length": 96.53125, + "epoch": 0.4748247291268324, + "grad_norm": 12.576179504394531, + "kl": 0.1845703125, + "learning_rate": 5.251752708731676e-07, + "loss": 0.0074, + "reward": 1.458923578262329, + "reward_std": 0.04576413705945015, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4589235782623291, + "rewards/pad": 0.0, + "step": 1490 + }, + { + "completion_length": 96.203125, + "epoch": 0.4751434034416826, + "grad_norm": 42.55897521972656, + "kl": 0.10693359375, + "learning_rate": 5.248565965583174e-07, + "loss": 0.0043, + "reward": 1.6549313068389893, + "reward_std": 0.03176398202776909, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5299313068389893, + "rewards/pad": 0.125, + "step": 1491 + }, + { + "completion_length": 70.375, + "epoch": 0.4754620777565328, + "grad_norm": 30.84855842590332, + "kl": 0.1787109375, + "learning_rate": 5.245379222434672e-07, + "loss": 0.0072, + "reward": 1.458336591720581, + "reward_std": 0.058646999299526215, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4583367109298706, + "rewards/pad": 0.0, + "step": 1492 + }, + { + "completion_length": 123.515625, + "epoch": 0.47578075207138304, + "grad_norm": 25.724525451660156, + "kl": 0.12890625, + "learning_rate": 5.24219247928617e-07, + "loss": 0.0052, + "reward": 1.5070421695709229, + "reward_std": 0.042530253529548645, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3820420503616333, + "rewards/pad": 0.125, + "step": 1493 + }, + { + "completion_length": 124.78125, + "epoch": 0.47609942638623326, + "grad_norm": 8.826605796813965, + "kl": 0.166015625, + "learning_rate": 5.239005736137667e-07, + "loss": 0.0066, + "reward": 1.6858264207839966, + "reward_std": 0.08236081898212433, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5608264207839966, + "step": 1494 + }, + { + "completion_length": 74.09375, + "epoch": 0.4764181007010835, + "grad_norm": 23.001249313354492, + "kl": 0.201171875, + "learning_rate": 5.235818992989165e-07, + "loss": 0.008, + "reward": 1.9178105592727661, + "reward_std": 0.060972314327955246, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.6678106188774109, + "step": 1495 + }, + { + "completion_length": 45.21875, + "epoch": 0.4767367750159337, + "grad_norm": 66.9561538696289, + "kl": 0.2275390625, + "learning_rate": 5.232632249840663e-07, + "loss": 0.0091, + "reward": 1.6306493282318115, + "reward_std": 0.15838010609149933, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5212743282318115, + "rewards/pad": 0.109375, + "step": 1496 + }, + { + "completion_length": 122.734375, + "epoch": 0.4770554493307839, + "grad_norm": 137.2368927001953, + "kl": 0.2158203125, + "learning_rate": 5.229445506692161e-07, + "loss": 0.0086, + "reward": 1.4414557218551636, + "reward_std": 0.11607854813337326, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4414557218551636, + "rewards/pad": 0.0, + "step": 1497 + }, + { + "completion_length": 44.546875, + "epoch": 0.47737412364563414, + "grad_norm": 26.18069839477539, + "kl": 0.169921875, + "learning_rate": 5.226258763543658e-07, + "loss": 0.0068, + "reward": 1.690140724182129, + "reward_std": 0.06070908159017563, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6901407241821289, + "rewards/pad": 0.0, + "step": 1498 + }, + { + "completion_length": 123.140625, + "epoch": 0.47769279796048436, + "grad_norm": 51.89798355102539, + "kl": 0.154296875, + "learning_rate": 5.223072020395156e-07, + "loss": 0.0062, + "reward": 1.5114344358444214, + "reward_std": 0.10912653058767319, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5114343166351318, + "rewards/pad": 0.0, + "step": 1499 + }, + { + "completion_length": 71.234375, + "epoch": 0.4780114722753346, + "grad_norm": 20.26959991455078, + "kl": 0.1259765625, + "learning_rate": 5.219885277246654e-07, + "loss": 0.005, + "reward": 1.6750046014785767, + "reward_std": 0.06571859866380692, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6750045418739319, + "rewards/pad": 0.0, + "step": 1500 + }, + { + "completion_length": 123.328125, + "epoch": 0.47833014659018486, + "grad_norm": 34.22078323364258, + "kl": 0.294921875, + "learning_rate": 5.216698534098152e-07, + "loss": 0.0118, + "reward": 1.624065637588501, + "reward_std": 0.07414637506008148, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.49906566739082336, + "rewards/pad": 0.125, + "step": 1501 + }, + { + "completion_length": 99.171875, + "epoch": 0.4786488209050351, + "grad_norm": 230.13027954101562, + "kl": 0.12451171875, + "learning_rate": 5.213511790949649e-07, + "loss": 0.005, + "reward": 1.460237979888916, + "reward_std": 0.08080984652042389, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.33523792028427124, + "step": 1502 + }, + { + "completion_length": 96.265625, + "epoch": 0.4789674952198853, + "grad_norm": 45.79084777832031, + "kl": 0.13671875, + "learning_rate": 5.210325047801147e-07, + "loss": 0.0055, + "reward": 1.4750096797943115, + "reward_std": 0.05699557065963745, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4750097692012787, + "rewards/pad": 0.0, + "step": 1503 + }, + { + "completion_length": 96.359375, + "epoch": 0.4792861695347355, + "grad_norm": 37.625885009765625, + "kl": 0.1806640625, + "learning_rate": 5.207138304652645e-07, + "loss": 0.0072, + "reward": 1.4821243286132812, + "reward_std": 0.060712795704603195, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4821242094039917, + "rewards/pad": 0.0, + "step": 1504 + }, + { + "completion_length": 95.40625, + "epoch": 0.47960484384958574, + "grad_norm": 45.98420333862305, + "kl": 0.193359375, + "learning_rate": 5.203951561504143e-07, + "loss": 0.0078, + "reward": 1.55684494972229, + "reward_std": 0.061466366052627563, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5568448901176453, + "step": 1505 + }, + { + "completion_length": 120.171875, + "epoch": 0.47992351816443596, + "grad_norm": 33.786048889160156, + "kl": 0.1748046875, + "learning_rate": 5.20076481835564e-07, + "loss": 0.007, + "reward": 1.4618111848831177, + "reward_std": 0.08361318707466125, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4618111550807953, + "step": 1506 + }, + { + "completion_length": 123.875, + "epoch": 0.4802421924792862, + "grad_norm": 30.181888580322266, + "kl": 0.126953125, + "learning_rate": 5.197578075207139e-07, + "loss": 0.0051, + "reward": 1.3651647567749023, + "reward_std": 0.060573190450668335, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.36516469717025757, + "step": 1507 + }, + { + "completion_length": 73.5625, + "epoch": 0.4805608667941364, + "grad_norm": 18.62019157409668, + "kl": 0.1552734375, + "learning_rate": 5.194391332058637e-07, + "loss": 0.0062, + "reward": 1.5799176692962646, + "reward_std": 0.06475313007831573, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.32991766929626465, + "step": 1508 + }, + { + "completion_length": 97.6875, + "epoch": 0.4808795411089866, + "grad_norm": 28.941173553466797, + "kl": 0.15234375, + "learning_rate": 5.191204588910135e-07, + "loss": 0.0061, + "reward": 1.4807729721069336, + "reward_std": 0.09656580537557602, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4807729125022888, + "step": 1509 + }, + { + "completion_length": 124.5625, + "epoch": 0.48119821542383684, + "grad_norm": 53.21528625488281, + "kl": 0.154296875, + "learning_rate": 5.188017845761632e-07, + "loss": 0.0062, + "reward": 1.8000701665878296, + "reward_std": 0.2111843079328537, + "rewards/format_reward_tg": 0.96875, + "rewards/iou_timestamp_reward": 0.4563201069831848, + "rewards/pad": 0.375, + "step": 1510 + }, + { + "completion_length": 71.765625, + "epoch": 0.48151688973868706, + "grad_norm": 11.812533378601074, + "kl": 0.21875, + "learning_rate": 5.184831102613129e-07, + "loss": 0.0088, + "reward": 1.581697702407837, + "reward_std": 0.09734717756509781, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5816976428031921, + "step": 1511 + }, + { + "completion_length": 98.25, + "epoch": 0.4818355640535373, + "grad_norm": 41.84068298339844, + "kl": 0.11376953125, + "learning_rate": 5.181644359464627e-07, + "loss": 0.0046, + "reward": 1.617072582244873, + "reward_std": 0.05539502203464508, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4920726418495178, + "step": 1512 + }, + { + "completion_length": 99.046875, + "epoch": 0.4821542383683875, + "grad_norm": 13.822397232055664, + "kl": 0.14453125, + "learning_rate": 5.178457616316124e-07, + "loss": 0.0058, + "reward": 1.7891383171081543, + "reward_std": 0.0863952562212944, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5391384363174438, + "step": 1513 + }, + { + "completion_length": 98.546875, + "epoch": 0.4824729126832377, + "grad_norm": 74.7471923828125, + "kl": 0.1455078125, + "learning_rate": 5.175270873167622e-07, + "loss": 0.0058, + "reward": 1.8373247385025024, + "reward_std": 0.10276351124048233, + "rewards/pad": 0.234375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.6029497385025024, + "step": 1514 + }, + { + "completion_length": 96.125, + "epoch": 0.48279158699808794, + "grad_norm": 23.02245330810547, + "kl": 0.138671875, + "learning_rate": 5.17208413001912e-07, + "loss": 0.0056, + "reward": 1.6412444114685059, + "reward_std": 0.0655876025557518, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5162444114685059, + "step": 1515 + }, + { + "completion_length": 150.0, + "epoch": 0.48311026131293816, + "grad_norm": 9.277081489562988, + "kl": 0.0908203125, + "learning_rate": 5.168897386870618e-07, + "loss": 0.0036, + "reward": 1.4091484546661377, + "reward_std": 0.04507271200418472, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4091483950614929, + "rewards/pad": 0.0, + "step": 1516 + }, + { + "completion_length": 71.6875, + "epoch": 0.4834289356277884, + "grad_norm": 32.54093551635742, + "kl": 0.220703125, + "learning_rate": 5.165710643722115e-07, + "loss": 0.0088, + "reward": 1.6252309083938599, + "reward_std": 0.1263033002614975, + "rewards/answer_reward": 0.109375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5158559083938599, + "step": 1517 + }, + { + "completion_length": 72.375, + "epoch": 0.4837476099426386, + "grad_norm": 57.833587646484375, + "kl": 0.1591796875, + "learning_rate": 5.162523900573613e-07, + "loss": 0.0064, + "reward": 1.628098964691162, + "reward_std": 0.10794929414987564, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5030990242958069, + "rewards/pad": 0.125, + "step": 1518 + }, + { + "completion_length": 49.78125, + "epoch": 0.4840662842574888, + "grad_norm": 35.0426139831543, + "kl": 0.259765625, + "learning_rate": 5.159337157425111e-07, + "loss": 0.0104, + "reward": 1.478843331336975, + "reward_std": 0.1224755346775055, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4319683313369751, + "rewards/pad": 0.046875, + "step": 1519 + }, + { + "completion_length": 96.265625, + "epoch": 0.4843849585723391, + "grad_norm": 33.974422454833984, + "kl": 0.10400390625, + "learning_rate": 5.156150414276609e-07, + "loss": 0.0042, + "reward": 1.6537295579910278, + "reward_std": 0.15965929627418518, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.5443546772003174, + "rewards/pad": 0.125, + "step": 1520 + }, + { + "completion_length": 99.453125, + "epoch": 0.4847036328871893, + "grad_norm": 24.93669891357422, + "kl": 0.2421875, + "learning_rate": 5.152963671128106e-07, + "loss": 0.0097, + "reward": 1.5643985271453857, + "reward_std": 0.13921087980270386, + "rewards/pad": 0.078125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.48627355694770813, + "step": 1521 + }, + { + "completion_length": 97.8125, + "epoch": 0.48502230720203954, + "grad_norm": 36.32746124267578, + "kl": 0.267578125, + "learning_rate": 5.149776927979604e-07, + "loss": 0.0107, + "reward": 1.6686410903930664, + "reward_std": 0.11134155094623566, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4342661499977112, + "rewards/pad": 0.234375, + "step": 1522 + }, + { + "completion_length": 123.609375, + "epoch": 0.48534098151688976, + "grad_norm": 33.87176513671875, + "kl": 0.14453125, + "learning_rate": 5.146590184831102e-07, + "loss": 0.0058, + "reward": 1.415372371673584, + "reward_std": 0.1284211277961731, + "rewards/pad": 0.03125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3841223120689392, + "step": 1523 + }, + { + "completion_length": 121.09375, + "epoch": 0.48565965583174, + "grad_norm": 46.95094680786133, + "kl": 0.1806640625, + "learning_rate": 5.1434034416826e-07, + "loss": 0.0072, + "reward": 1.332892894744873, + "reward_std": 0.05756673216819763, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.33289292454719543, + "step": 1524 + }, + { + "completion_length": 175.265625, + "epoch": 0.4859783301465902, + "grad_norm": 11.277741432189941, + "kl": 0.1376953125, + "learning_rate": 5.140216698534097e-07, + "loss": 0.0055, + "reward": 1.375731110572815, + "reward_std": 0.0918242484331131, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.3913561999797821, + "step": 1525 + }, + { + "completion_length": 72.609375, + "epoch": 0.4862970044614404, + "grad_norm": 29.35453224182129, + "kl": 0.3828125, + "learning_rate": 5.137029955385595e-07, + "loss": 0.0153, + "reward": 1.5956140756607056, + "reward_std": 0.07637912034988403, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.3456140160560608, + "step": 1526 + }, + { + "completion_length": 70.859375, + "epoch": 0.48661567877629064, + "grad_norm": 27.634729385375977, + "kl": 0.6484375, + "learning_rate": 5.133843212237094e-07, + "loss": 0.0259, + "reward": 1.6848344802856445, + "reward_std": 0.07616157829761505, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.559834361076355, + "rewards/pad": 0.125, + "step": 1527 + }, + { + "completion_length": 19.9375, + "epoch": 0.48693435309114086, + "grad_norm": 43.502281188964844, + "kl": 0.1943359375, + "learning_rate": 5.130656469088592e-07, + "loss": 0.0078, + "reward": 1.6018435955047607, + "reward_std": 0.09998296201229095, + "rewards/answer_reward": 0.15625, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.44559359550476074, + "step": 1528 + }, + { + "completion_length": 44.546875, + "epoch": 0.4872530274059911, + "grad_norm": 28.72425651550293, + "kl": 0.169921875, + "learning_rate": 5.127469725940089e-07, + "loss": 0.0068, + "reward": 1.4866104125976562, + "reward_std": 0.05266030877828598, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.486610472202301, + "rewards/pad": 0.0, + "step": 1529 + }, + { + "completion_length": 47.125, + "epoch": 0.4875717017208413, + "grad_norm": 71.66815185546875, + "kl": 0.2275390625, + "learning_rate": 5.124282982791587e-07, + "loss": 0.0091, + "reward": 1.5927340984344482, + "reward_std": 0.16904056072235107, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.37398403882980347, + "rewards/pad": 0.234375, + "step": 1530 + }, + { + "completion_length": 72.515625, + "epoch": 0.4878903760356915, + "grad_norm": 198.94308471679688, + "kl": 0.1845703125, + "learning_rate": 5.121096239643085e-07, + "loss": 0.0074, + "reward": 1.823049545288086, + "reward_std": 0.13386155664920807, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6355496644973755, + "rewards/pad": 0.1875, + "step": 1531 + }, + { + "completion_length": 72.171875, + "epoch": 0.48820905035054174, + "grad_norm": 25.96636962890625, + "kl": 0.158203125, + "learning_rate": 5.117909496494583e-07, + "loss": 0.0063, + "reward": 1.523979902267456, + "reward_std": 0.059931959956884384, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.523979902267456, + "rewards/pad": 0.0, + "step": 1532 + }, + { + "completion_length": 71.3125, + "epoch": 0.48852772466539196, + "grad_norm": 38.72029495239258, + "kl": 0.18359375, + "learning_rate": 5.11472275334608e-07, + "loss": 0.0073, + "reward": 1.4667209386825562, + "reward_std": 0.09433542937040329, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4510958790779114, + "rewards/pad": 0.015625, + "step": 1533 + }, + { + "completion_length": 72.09375, + "epoch": 0.4888463989802422, + "grad_norm": 86.83837890625, + "kl": 0.1904296875, + "learning_rate": 5.111536010197578e-07, + "loss": 0.0076, + "reward": 1.7294871807098389, + "reward_std": 0.16661599278450012, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.49511218070983887, + "rewards/pad": 0.25, + "step": 1534 + }, + { + "completion_length": 128.984375, + "epoch": 0.4891650732950924, + "grad_norm": 215.36953735351562, + "kl": 0.1220703125, + "learning_rate": 5.108349267049076e-07, + "loss": 0.0049, + "reward": 1.614091396331787, + "reward_std": 0.08909813314676285, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.3640914559364319, + "step": 1535 + }, + { + "completion_length": 149.21875, + "epoch": 0.4894837476099426, + "grad_norm": 5.865546703338623, + "kl": 0.11572265625, + "learning_rate": 5.105162523900574e-07, + "loss": 0.0046, + "reward": 1.529909372329712, + "reward_std": 0.040073707699775696, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4049093723297119, + "rewards/pad": 0.125, + "step": 1536 + }, + { + "completion_length": 122.84375, + "epoch": 0.48980242192479284, + "grad_norm": 22.46348762512207, + "kl": 0.1162109375, + "learning_rate": 5.101975780752071e-07, + "loss": 0.0046, + "reward": 1.4626240730285645, + "reward_std": 0.095807746052742, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.4782490134239197, + "step": 1537 + }, + { + "completion_length": 77.125, + "epoch": 0.49012109623964306, + "grad_norm": 45.38951110839844, + "kl": 0.158203125, + "learning_rate": 5.098789037603569e-07, + "loss": 0.0063, + "reward": 1.5587177276611328, + "reward_std": 0.08830150961875916, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4337177276611328, + "step": 1538 + }, + { + "completion_length": 98.390625, + "epoch": 0.4904397705544933, + "grad_norm": 29.60246467590332, + "kl": 0.1953125, + "learning_rate": 5.095602294455067e-07, + "loss": 0.0078, + "reward": 1.63431978225708, + "reward_std": 0.06482817977666855, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.6343198418617249, + "step": 1539 + }, + { + "completion_length": 46.0, + "epoch": 0.49075844486934356, + "grad_norm": 24.609296798706055, + "kl": 0.1826171875, + "learning_rate": 5.092415551306564e-07, + "loss": 0.0073, + "reward": 1.4860643148422241, + "reward_std": 0.09358102083206177, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.47043925523757935, + "rewards/pad": 0.015625, + "step": 1540 + }, + { + "completion_length": 99.0625, + "epoch": 0.4910771191841938, + "grad_norm": 23.453184127807617, + "kl": 0.11181640625, + "learning_rate": 5.089228808158062e-07, + "loss": 0.0045, + "reward": 1.7013800144195557, + "reward_std": 0.09791174530982971, + "rewards/pad": 0.109375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5920050144195557, + "step": 1541 + }, + { + "completion_length": 123.6875, + "epoch": 0.491395793499044, + "grad_norm": 21.338483810424805, + "kl": 0.1767578125, + "learning_rate": 5.08604206500956e-07, + "loss": 0.0071, + "reward": 1.5143307447433472, + "reward_std": 0.0737447589635849, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.38933080434799194, + "step": 1542 + }, + { + "completion_length": 78.4375, + "epoch": 0.4917144678138942, + "grad_norm": 34.1049919128418, + "kl": 0.224609375, + "learning_rate": 5.082855321861058e-07, + "loss": 0.009, + "reward": 1.594736933708191, + "reward_std": 0.08642074465751648, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5947369337081909, + "rewards/pad": 0.0, + "step": 1543 + }, + { + "completion_length": 75.484375, + "epoch": 0.49203314212874444, + "grad_norm": 35.719322204589844, + "kl": 0.26953125, + "learning_rate": 5.079668578712555e-07, + "loss": 0.0108, + "reward": 1.5955408811569214, + "reward_std": 0.16018131375312805, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4861658215522766, + "rewards/pad": 0.109375, + "step": 1544 + }, + { + "completion_length": 44.703125, + "epoch": 0.49235181644359466, + "grad_norm": 30.363176345825195, + "kl": 0.298828125, + "learning_rate": 5.076481835564054e-07, + "loss": 0.012, + "reward": 1.5528295040130615, + "reward_std": 0.08602527529001236, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5528294444084167, + "step": 1545 + }, + { + "completion_length": 71.640625, + "epoch": 0.4926704907584449, + "grad_norm": 34.98065185546875, + "kl": 0.1181640625, + "learning_rate": 5.073295092415552e-07, + "loss": 0.0047, + "reward": 1.6527106761932373, + "reward_std": 0.04381757974624634, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5277107357978821, + "step": 1546 + }, + { + "completion_length": 93.40625, + "epoch": 0.4929891650732951, + "grad_norm": 13.550950050354004, + "kl": 0.1826171875, + "learning_rate": 5.07010834926705e-07, + "loss": 0.0073, + "reward": 1.5585880279541016, + "reward_std": 0.08185498416423798, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5585879683494568, + "step": 1547 + }, + { + "completion_length": 96.34375, + "epoch": 0.4933078393881453, + "grad_norm": 25.857145309448242, + "kl": 0.1142578125, + "learning_rate": 5.066921606118547e-07, + "loss": 0.0046, + "reward": 1.4285244941711426, + "reward_std": 0.1266767680644989, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.44414952397346497, + "rewards/pad": 0.0, + "step": 1548 + }, + { + "completion_length": 72.859375, + "epoch": 0.49362651370299554, + "grad_norm": 42.86756896972656, + "kl": 0.18359375, + "learning_rate": 5.063734862970045e-07, + "loss": 0.0074, + "reward": 1.8299238681793213, + "reward_std": 0.16717174649238586, + "rewards/answer_reward": 0.390625, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.43929874897003174, + "step": 1549 + }, + { + "completion_length": 149.015625, + "epoch": 0.49394518801784576, + "grad_norm": 29.9815731048584, + "kl": 0.1357421875, + "learning_rate": 5.060548119821542e-07, + "loss": 0.0054, + "reward": 1.3113267421722412, + "reward_std": 0.0599776916205883, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3113267123699188, + "step": 1550 + }, + { + "completion_length": 70.734375, + "epoch": 0.494263862332696, + "grad_norm": 38.25667953491211, + "kl": 0.173828125, + "learning_rate": 5.05736137667304e-07, + "loss": 0.0069, + "reward": 1.47730553150177, + "reward_std": 0.0800052136182785, + "rewards/answer_reward": 0.0, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.47730547189712524, + "step": 1551 + }, + { + "completion_length": 123.984375, + "epoch": 0.4945825366475462, + "grad_norm": 52.59759521484375, + "kl": 0.1318359375, + "learning_rate": 5.054174633524537e-07, + "loss": 0.0052, + "reward": 1.4347107410430908, + "reward_std": 0.12221598625183105, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.45033565163612366, + "step": 1552 + }, + { + "completion_length": 97.390625, + "epoch": 0.4949012109623964, + "grad_norm": 55.0565185546875, + "kl": 0.1298828125, + "learning_rate": 5.050987890376035e-07, + "loss": 0.0052, + "reward": 1.5172079801559448, + "reward_std": 0.09507996588945389, + "rewards/pad": 0.015625, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5015830397605896, + "step": 1553 + }, + { + "completion_length": 73.265625, + "epoch": 0.49521988527724664, + "grad_norm": 55.56892776489258, + "kl": 0.1923828125, + "learning_rate": 5.047801147227533e-07, + "loss": 0.0077, + "reward": 1.5761849880218506, + "reward_std": 0.09632700681686401, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.45118507742881775, + "step": 1554 + }, + { + "completion_length": 93.921875, + "epoch": 0.49553855959209686, + "grad_norm": 18.484773635864258, + "kl": 0.1494140625, + "learning_rate": 5.044614404079031e-07, + "loss": 0.006, + "reward": 1.4626483917236328, + "reward_std": 0.13790366053581238, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.4782733619213104, + "rewards/pad": 0.0, + "step": 1555 + }, + { + "completion_length": 69.1875, + "epoch": 0.4958572339069471, + "grad_norm": 22.0714111328125, + "kl": 0.236328125, + "learning_rate": 5.041427660930528e-07, + "loss": 0.0094, + "reward": 1.6114702224731445, + "reward_std": 0.058457039296627045, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6114701628684998, + "rewards/pad": 0.0, + "step": 1556 + }, + { + "completion_length": 146.265625, + "epoch": 0.4961759082217973, + "grad_norm": 16.967422485351562, + "kl": 0.08154296875, + "learning_rate": 5.038240917782026e-07, + "loss": 0.0033, + "reward": 1.2791426181793213, + "reward_std": 0.02925313636660576, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.2791425287723541, + "rewards/pad": 0.0, + "step": 1557 + }, + { + "completion_length": 96.34375, + "epoch": 0.4964945825366475, + "grad_norm": 135.3175506591797, + "kl": 0.1572265625, + "learning_rate": 5.035054174633524e-07, + "loss": 0.0063, + "reward": 1.4358652830123901, + "reward_std": 0.03743431344628334, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.43586528301239014, + "rewards/pad": 0.0, + "step": 1558 + }, + { + "completion_length": 97.671875, + "epoch": 0.49681325685149774, + "grad_norm": 105.35767364501953, + "kl": 0.138671875, + "learning_rate": 5.031867431485022e-07, + "loss": 0.0055, + "reward": 1.5372705459594727, + "reward_std": 0.04880628362298012, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4122704565525055, + "step": 1559 + }, + { + "completion_length": 93.53125, + "epoch": 0.497131931166348, + "grad_norm": 21.049678802490234, + "kl": 0.15234375, + "learning_rate": 5.028680688336519e-07, + "loss": 0.0061, + "reward": 1.59987211227417, + "reward_std": 0.03113095834851265, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5998720526695251, + "rewards/pad": 0.0, + "step": 1560 + }, + { + "completion_length": 46.875, + "epoch": 0.49745060548119824, + "grad_norm": 48.804405212402344, + "kl": 0.25390625, + "learning_rate": 5.025493945188017e-07, + "loss": 0.0102, + "reward": 1.4484260082244873, + "reward_std": 0.10934975743293762, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4328010678291321, + "rewards/pad": 0.015625, + "step": 1561 + }, + { + "completion_length": 48.59375, + "epoch": 0.49776927979604846, + "grad_norm": 56.09344482421875, + "kl": 0.18359375, + "learning_rate": 5.022307202039515e-07, + "loss": 0.0073, + "reward": 1.8689539432525635, + "reward_std": 0.12545578181743622, + "rewards/answer_reward": 0.421875, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4470789134502411, + "step": 1562 + }, + { + "completion_length": 70.5, + "epoch": 0.4980879541108987, + "grad_norm": 47.180362701416016, + "kl": 0.1904296875, + "learning_rate": 5.019120458891013e-07, + "loss": 0.0076, + "reward": 1.495100736618042, + "reward_std": 0.0501336008310318, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4951007068157196, + "step": 1563 + }, + { + "completion_length": 170.34375, + "epoch": 0.4984066284257489, + "grad_norm": 19.715389251708984, + "kl": 0.06494140625, + "learning_rate": 5.01593371574251e-07, + "loss": 0.0026, + "reward": 1.3467738628387451, + "reward_std": 0.028132686391472816, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3467738628387451, + "step": 1564 + }, + { + "completion_length": 72.390625, + "epoch": 0.4987253027405991, + "grad_norm": 39.21231460571289, + "kl": 0.1298828125, + "learning_rate": 5.012746972594009e-07, + "loss": 0.0052, + "reward": 1.3724713325500488, + "reward_std": 0.07831380516290665, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3568463623523712, + "rewards/pad": 0.015625, + "step": 1565 + }, + { + "completion_length": 97.78125, + "epoch": 0.49904397705544934, + "grad_norm": 70.9393310546875, + "kl": 0.181640625, + "learning_rate": 5.009560229445507e-07, + "loss": 0.0073, + "reward": 1.560790777206421, + "reward_std": 0.04791503772139549, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4357907772064209, + "rewards/pad": 0.125, + "step": 1566 + }, + { + "completion_length": 73.90625, + "epoch": 0.49936265137029956, + "grad_norm": 56.073753356933594, + "kl": 0.13671875, + "learning_rate": 5.006373486297005e-07, + "loss": 0.0055, + "reward": 1.8846735954284668, + "reward_std": 0.08357321470975876, + "rewards/answer_reward": 0.359375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5252986550331116, + "step": 1567 + }, + { + "completion_length": 96.46875, + "epoch": 0.4996813256851498, + "grad_norm": 22.90010643005371, + "kl": 0.130859375, + "learning_rate": 5.003186743148502e-07, + "loss": 0.0052, + "reward": 1.486340045928955, + "reward_std": 0.05758453905582428, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4863400161266327, + "step": 1568 + }, + { + "completion_length": 96.0625, + "epoch": 0.5, + "grad_norm": 109.51544952392578, + "kl": 0.1513671875, + "learning_rate": 5e-07, + "loss": 0.0061, + "reward": 1.646225094795227, + "reward_std": 0.10367502272129059, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5212252140045166, + "step": 1569 + }, + { + "completion_length": 95.65625, + "epoch": 0.5003186743148502, + "grad_norm": 18.478254318237305, + "kl": 0.1455078125, + "learning_rate": 4.996813256851498e-07, + "loss": 0.0058, + "reward": 1.5283114910125732, + "reward_std": 0.0567050576210022, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.528311550617218, + "rewards/pad": 0.0, + "step": 1570 + }, + { + "completion_length": 72.8125, + "epoch": 0.5006373486297004, + "grad_norm": 40.840431213378906, + "kl": 0.263671875, + "learning_rate": 4.993626513702995e-07, + "loss": 0.0105, + "reward": 1.5773298740386963, + "reward_std": 0.1823175549507141, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4835800230503082, + "rewards/pad": 0.09375, + "step": 1571 + }, + { + "completion_length": 120.953125, + "epoch": 0.5009560229445507, + "grad_norm": 69.60863494873047, + "kl": 0.0986328125, + "learning_rate": 4.990439770554493e-07, + "loss": 0.0039, + "reward": 1.693233609199524, + "reward_std": 0.07272499799728394, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5682336091995239, + "rewards/pad": 0.125, + "step": 1572 + }, + { + "completion_length": 96.765625, + "epoch": 0.5012746972594009, + "grad_norm": 42.9666748046875, + "kl": 0.158203125, + "learning_rate": 4.987253027405991e-07, + "loss": 0.0063, + "reward": 1.7116810083389282, + "reward_std": 0.09147972613573074, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.586681067943573, + "step": 1573 + }, + { + "completion_length": 70.09375, + "epoch": 0.5015933715742511, + "grad_norm": 39.15825271606445, + "kl": 0.23046875, + "learning_rate": 4.984066284257489e-07, + "loss": 0.0092, + "reward": 1.6254063844680786, + "reward_std": 0.10068808495998383, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5004063844680786, + "rewards/pad": 0.125, + "step": 1574 + }, + { + "completion_length": 95.953125, + "epoch": 0.5019120458891013, + "grad_norm": 2402.987060546875, + "kl": 5.6875, + "learning_rate": 4.980879541108986e-07, + "loss": 0.227, + "reward": 1.4909906387329102, + "reward_std": 0.1678842306137085, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.5066156387329102, + "step": 1575 + }, + { + "completion_length": 94.6875, + "epoch": 0.5022307202039515, + "grad_norm": 21.808799743652344, + "kl": 0.11962890625, + "learning_rate": 4.977692797960484e-07, + "loss": 0.0048, + "reward": 1.6425373554229736, + "reward_std": 0.14815226197242737, + "rewards/pad": 0.171875, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4706624746322632, + "step": 1576 + }, + { + "completion_length": 70.484375, + "epoch": 0.5025493945188018, + "grad_norm": 24.197406768798828, + "kl": 0.1455078125, + "learning_rate": 4.974506054811982e-07, + "loss": 0.0058, + "reward": 1.5822218656539917, + "reward_std": 0.059086866676807404, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5822218656539917, + "step": 1577 + }, + { + "completion_length": 124.546875, + "epoch": 0.502868068833652, + "grad_norm": 59.101585388183594, + "kl": 0.193359375, + "learning_rate": 4.97131931166348e-07, + "loss": 0.0077, + "reward": 1.5973970890045166, + "reward_std": 0.08404646068811417, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5973970890045166, + "step": 1578 + }, + { + "completion_length": 96.203125, + "epoch": 0.5031867431485022, + "grad_norm": 16.71595573425293, + "kl": 0.126953125, + "learning_rate": 4.968132568514977e-07, + "loss": 0.0051, + "reward": 1.449321985244751, + "reward_std": 0.05522039160132408, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.44932207465171814, + "step": 1579 + }, + { + "completion_length": 98.015625, + "epoch": 0.5035054174633524, + "grad_norm": 24.666296005249023, + "kl": 0.08984375, + "learning_rate": 4.964945825366475e-07, + "loss": 0.0036, + "reward": 1.659955620765686, + "reward_std": 0.07834160327911377, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4099557101726532, + "step": 1580 + }, + { + "completion_length": 145.28125, + "epoch": 0.5038240917782026, + "grad_norm": 16.429119110107422, + "kl": 0.12060546875, + "learning_rate": 4.961759082217972e-07, + "loss": 0.0048, + "reward": 1.436887264251709, + "reward_std": 0.0728699266910553, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4368871748447418, + "step": 1581 + }, + { + "completion_length": 72.90625, + "epoch": 0.5041427660930529, + "grad_norm": 35.17189407348633, + "kl": 0.15625, + "learning_rate": 4.95857233906947e-07, + "loss": 0.0062, + "reward": 1.5288763046264648, + "reward_std": 0.1308014839887619, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.5445013046264648, + "step": 1582 + }, + { + "completion_length": 71.140625, + "epoch": 0.5044614404079031, + "grad_norm": 54.288818359375, + "kl": 0.1494140625, + "learning_rate": 4.955385595920969e-07, + "loss": 0.006, + "reward": 1.511135458946228, + "reward_std": 0.06047463044524193, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5111355185508728, + "step": 1583 + }, + { + "completion_length": 149.578125, + "epoch": 0.5047801147227533, + "grad_norm": 10.70415210723877, + "kl": 0.09033203125, + "learning_rate": 4.952198852772467e-07, + "loss": 0.0036, + "reward": 1.6119954586029053, + "reward_std": 0.04025973752140999, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3619953691959381, + "step": 1584 + }, + { + "completion_length": 73.015625, + "epoch": 0.5050987890376035, + "grad_norm": 39.62226104736328, + "kl": 0.1279296875, + "learning_rate": 4.949012109623964e-07, + "loss": 0.0051, + "reward": 1.833071231842041, + "reward_std": 0.11723049730062485, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5361962914466858, + "rewards/pad": 0.296875, + "step": 1585 + }, + { + "completion_length": 96.328125, + "epoch": 0.5054174633524537, + "grad_norm": 9.883891105651855, + "kl": 0.2353515625, + "learning_rate": 4.945825366475462e-07, + "loss": 0.0094, + "reward": 1.5089480876922607, + "reward_std": 0.0629185289144516, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5089481472969055, + "rewards/pad": 0.0, + "step": 1586 + }, + { + "completion_length": 68.953125, + "epoch": 0.505736137667304, + "grad_norm": 41.82442855834961, + "kl": 0.162109375, + "learning_rate": 4.94263862332696e-07, + "loss": 0.0065, + "reward": 1.6769931316375732, + "reward_std": 0.08155511319637299, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5519931316375732, + "step": 1587 + }, + { + "completion_length": 147.109375, + "epoch": 0.5060548119821542, + "grad_norm": 64.26282501220703, + "kl": 0.09619140625, + "learning_rate": 4.939451880178458e-07, + "loss": 0.0038, + "reward": 1.5831506252288818, + "reward_std": 0.11183653771877289, + "rewards/pad": 0.109375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.47377562522888184, + "step": 1588 + }, + { + "completion_length": 147.46875, + "epoch": 0.5063734862970045, + "grad_norm": 9.992218971252441, + "kl": 0.1142578125, + "learning_rate": 4.936265137029955e-07, + "loss": 0.0046, + "reward": 1.7075880765914917, + "reward_std": 0.10298748314380646, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5825880169868469, + "step": 1589 + }, + { + "completion_length": 126.546875, + "epoch": 0.5066921606118547, + "grad_norm": 98.55374145507812, + "kl": 0.2177734375, + "learning_rate": 4.933078393881453e-07, + "loss": 0.0087, + "reward": 1.611853837966919, + "reward_std": 0.03698485344648361, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4868537187576294, + "step": 1590 + }, + { + "completion_length": 43.859375, + "epoch": 0.507010834926705, + "grad_norm": 29.617713928222656, + "kl": 0.1337890625, + "learning_rate": 4.929891650732951e-07, + "loss": 0.0053, + "reward": 1.757079005241394, + "reward_std": 0.06662129610776901, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.6320791244506836, + "step": 1591 + }, + { + "completion_length": 69.359375, + "epoch": 0.5073295092415552, + "grad_norm": 51.22451400756836, + "kl": 0.2041015625, + "learning_rate": 4.926704907584449e-07, + "loss": 0.0082, + "reward": 1.728276252746582, + "reward_std": 0.06753535568714142, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.6032761931419373, + "step": 1592 + }, + { + "completion_length": 121.484375, + "epoch": 0.5076481835564054, + "grad_norm": 18.68769073486328, + "kl": 0.103515625, + "learning_rate": 4.923518164435946e-07, + "loss": 0.0041, + "reward": 1.4825172424316406, + "reward_std": 0.07809288054704666, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.35751718282699585, + "step": 1593 + }, + { + "completion_length": 123.265625, + "epoch": 0.5079668578712556, + "grad_norm": 35.22314453125, + "kl": 0.10400390625, + "learning_rate": 4.920331421287444e-07, + "loss": 0.0042, + "reward": 1.5061140060424805, + "reward_std": 0.04520394280552864, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.38111400604248047, + "rewards/pad": 0.125, + "step": 1594 + }, + { + "completion_length": 99.296875, + "epoch": 0.5082855321861058, + "grad_norm": 18.381580352783203, + "kl": 0.10546875, + "learning_rate": 4.917144678138942e-07, + "loss": 0.0042, + "reward": 1.8121449947357178, + "reward_std": 0.045704472810029984, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.43714502453804016, + "rewards/pad": 0.375, + "step": 1595 + }, + { + "completion_length": 94.046875, + "epoch": 0.5086042065009561, + "grad_norm": 26.083742141723633, + "kl": 0.1494140625, + "learning_rate": 4.91395793499044e-07, + "loss": 0.006, + "reward": 1.4526923894882202, + "reward_std": 0.06620436906814575, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4526923596858978, + "rewards/pad": 0.0, + "step": 1596 + }, + { + "completion_length": 148.984375, + "epoch": 0.5089228808158063, + "grad_norm": 55.55546951293945, + "kl": 0.1162109375, + "learning_rate": 4.910771191841937e-07, + "loss": 0.0047, + "reward": 1.3445849418640137, + "reward_std": 0.036867473274469376, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3445849120616913, + "step": 1597 + }, + { + "completion_length": 144.03125, + "epoch": 0.5092415551306565, + "grad_norm": 17.205900192260742, + "kl": 0.123046875, + "learning_rate": 4.907584448693435e-07, + "loss": 0.0049, + "reward": 1.506199836730957, + "reward_std": 0.05332973971962929, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5061998963356018, + "rewards/pad": 0.0, + "step": 1598 + }, + { + "completion_length": 95.921875, + "epoch": 0.5095602294455067, + "grad_norm": 21.30576515197754, + "kl": 0.20703125, + "learning_rate": 4.904397705544932e-07, + "loss": 0.0083, + "reward": 1.6577305793762207, + "reward_std": 0.11153335869312286, + "rewards/pad": 0.03125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.6264805793762207, + "step": 1599 + }, + { + "completion_length": 94.34375, + "epoch": 0.5098789037603569, + "grad_norm": 47.93109130859375, + "kl": 0.162109375, + "learning_rate": 4.90121096239643e-07, + "loss": 0.0065, + "reward": 1.5295488834381104, + "reward_std": 0.15334632992744446, + "rewards/pad": 0.03125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.49829891324043274, + "step": 1600 + }, + { + "completion_length": 71.984375, + "epoch": 0.5101975780752072, + "grad_norm": 46.93277359008789, + "kl": 0.244140625, + "learning_rate": 4.898024219247928e-07, + "loss": 0.0098, + "reward": 1.6092555522918701, + "reward_std": 0.07779088616371155, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.48425549268722534, + "rewards/pad": 0.125, + "step": 1601 + }, + { + "completion_length": 43.625, + "epoch": 0.5105162523900574, + "grad_norm": 47.682518005371094, + "kl": 0.2001953125, + "learning_rate": 4.894837476099425e-07, + "loss": 0.008, + "reward": 1.5027817487716675, + "reward_std": 0.06517623364925385, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5027817487716675, + "step": 1602 + }, + { + "completion_length": 123.015625, + "epoch": 0.5108349267049076, + "grad_norm": 58.71294403076172, + "kl": 0.0859375, + "learning_rate": 4.891650732950924e-07, + "loss": 0.0034, + "reward": 1.413274884223938, + "reward_std": 0.12114940583705902, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.319524884223938, + "rewards/pad": 0.09375, + "step": 1603 + }, + { + "completion_length": 120.390625, + "epoch": 0.5111536010197578, + "grad_norm": 10.46804428100586, + "kl": 0.142578125, + "learning_rate": 4.888463989802422e-07, + "loss": 0.0057, + "reward": 1.3477072715759277, + "reward_std": 0.06476961821317673, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3477073311805725, + "step": 1604 + }, + { + "completion_length": 96.28125, + "epoch": 0.511472275334608, + "grad_norm": 27.42978286743164, + "kl": 0.130859375, + "learning_rate": 4.88527724665392e-07, + "loss": 0.0052, + "reward": 1.3898663520812988, + "reward_std": 0.07068990170955658, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3898663818836212, + "rewards/pad": 0.0, + "step": 1605 + }, + { + "completion_length": 96.046875, + "epoch": 0.5117909496494583, + "grad_norm": 13.107078552246094, + "kl": 0.1533203125, + "learning_rate": 4.882090503505417e-07, + "loss": 0.0061, + "reward": 1.6023199558258057, + "reward_std": 0.052031271159648895, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4773198366165161, + "step": 1606 + }, + { + "completion_length": 123.078125, + "epoch": 0.5121096239643085, + "grad_norm": 27.67726707458496, + "kl": 0.1650390625, + "learning_rate": 4.878903760356915e-07, + "loss": 0.0066, + "reward": 1.6470005512237549, + "reward_std": 0.09419877827167511, + "rewards/answer_reward": 0.140625, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5063754320144653, + "step": 1607 + }, + { + "completion_length": 147.03125, + "epoch": 0.5124282982791587, + "grad_norm": 21.777843475341797, + "kl": 0.10107421875, + "learning_rate": 4.875717017208413e-07, + "loss": 0.004, + "reward": 1.5854337215423584, + "reward_std": 0.042907871305942535, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.585433840751648, + "step": 1608 + }, + { + "completion_length": 72.125, + "epoch": 0.5127469725940089, + "grad_norm": 37.6806526184082, + "kl": 0.173828125, + "learning_rate": 4.872530274059911e-07, + "loss": 0.0069, + "reward": 1.4564390182495117, + "reward_std": 0.10523393005132675, + "rewards/answer_reward": 0.015625, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4408140778541565, + "step": 1609 + }, + { + "completion_length": 95.5, + "epoch": 0.5130656469088591, + "grad_norm": 36.77358627319336, + "kl": 0.1611328125, + "learning_rate": 4.869343530911408e-07, + "loss": 0.0065, + "reward": 1.5982133150100708, + "reward_std": 0.09956274181604385, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5982133150100708, + "rewards/pad": 0.0, + "step": 1610 + }, + { + "completion_length": 45.125, + "epoch": 0.5133843212237094, + "grad_norm": 34.843727111816406, + "kl": 0.1767578125, + "learning_rate": 4.866156787762906e-07, + "loss": 0.0071, + "reward": 1.751644253730774, + "reward_std": 0.09237408638000488, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6266443133354187, + "rewards/pad": 0.125, + "step": 1611 + }, + { + "completion_length": 47.890625, + "epoch": 0.5137029955385596, + "grad_norm": 27.2742977142334, + "kl": 0.1474609375, + "learning_rate": 4.862970044614404e-07, + "loss": 0.0059, + "reward": 1.4951512813568115, + "reward_std": 0.0747571736574173, + "rewards/answer_reward": 0.0, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.49515125155448914, + "step": 1612 + }, + { + "completion_length": 79.53125, + "epoch": 0.5140216698534098, + "grad_norm": 72.64749908447266, + "kl": 0.123046875, + "learning_rate": 4.859783301465902e-07, + "loss": 0.0049, + "reward": 1.7619708776474, + "reward_std": 0.14130009710788727, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.4025958776473999, + "rewards/pad": 0.375, + "step": 1613 + }, + { + "completion_length": 145.65625, + "epoch": 0.51434034416826, + "grad_norm": 78.582763671875, + "kl": 0.10302734375, + "learning_rate": 4.856596558317399e-07, + "loss": 0.0041, + "reward": 1.4470350742340088, + "reward_std": 0.04571251943707466, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4470349848270416, + "rewards/pad": 0.0, + "step": 1614 + }, + { + "completion_length": 121.46875, + "epoch": 0.5146590184831102, + "grad_norm": 26.473241806030273, + "kl": 0.138671875, + "learning_rate": 4.853409815168897e-07, + "loss": 0.0055, + "reward": 1.5637502670288086, + "reward_std": 0.04434807598590851, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.43875014781951904, + "rewards/pad": 0.125, + "step": 1615 + }, + { + "completion_length": 175.34375, + "epoch": 0.5149776927979605, + "grad_norm": 26.083354949951172, + "kl": 0.06884765625, + "learning_rate": 4.850223072020395e-07, + "loss": 0.0028, + "reward": 1.5136511325836182, + "reward_std": 0.04185505956411362, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5136511325836182, + "step": 1616 + }, + { + "completion_length": 44.953125, + "epoch": 0.5152963671128107, + "grad_norm": 35.294132232666016, + "kl": 0.26171875, + "learning_rate": 4.847036328871893e-07, + "loss": 0.0104, + "reward": 1.489182710647583, + "reward_std": 0.12016036361455917, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.48918282985687256, + "rewards/pad": 0.0, + "step": 1617 + }, + { + "completion_length": 173.328125, + "epoch": 0.5156150414276609, + "grad_norm": 12.702577590942383, + "kl": 0.06884765625, + "learning_rate": 4.84384958572339e-07, + "loss": 0.0027, + "reward": 1.5943570137023926, + "reward_std": 0.05077844485640526, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4693570137023926, + "step": 1618 + }, + { + "completion_length": 148.140625, + "epoch": 0.5159337157425111, + "grad_norm": 28.387948989868164, + "kl": 0.11474609375, + "learning_rate": 4.840662842574888e-07, + "loss": 0.0046, + "reward": 1.4957976341247559, + "reward_std": 0.06421782076358795, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4957975745201111, + "rewards/pad": 0.0, + "step": 1619 + }, + { + "completion_length": 149.6875, + "epoch": 0.5162523900573613, + "grad_norm": 63.94855880737305, + "kl": 0.11181640625, + "learning_rate": 4.837476099426385e-07, + "loss": 0.0045, + "reward": 1.4858911037445068, + "reward_std": 0.05516339838504791, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.48589110374450684, + "step": 1620 + }, + { + "completion_length": 119.703125, + "epoch": 0.5165710643722116, + "grad_norm": 37.80052185058594, + "kl": 0.0751953125, + "learning_rate": 4.834289356277884e-07, + "loss": 0.003, + "reward": 1.5186381340026855, + "reward_std": 0.04086693376302719, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5186381340026855, + "step": 1621 + }, + { + "completion_length": 121.25, + "epoch": 0.5168897386870618, + "grad_norm": 9.314208030700684, + "kl": 0.1259765625, + "learning_rate": 4.831102613129382e-07, + "loss": 0.005, + "reward": 1.4766916036605835, + "reward_std": 0.03902985155582428, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4766915440559387, + "step": 1622 + }, + { + "completion_length": 98.828125, + "epoch": 0.517208413001912, + "grad_norm": 48.123897552490234, + "kl": 0.1162109375, + "learning_rate": 4.82791586998088e-07, + "loss": 0.0047, + "reward": 1.5715768337249756, + "reward_std": 0.06291652470827103, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.44657695293426514, + "rewards/pad": 0.125, + "step": 1623 + }, + { + "completion_length": 20.765625, + "epoch": 0.5175270873167622, + "grad_norm": 43.335289001464844, + "kl": 0.142578125, + "learning_rate": 4.824729126832377e-07, + "loss": 0.0057, + "reward": 1.7628185749053955, + "reward_std": 0.20459821820259094, + "rewards/answer_reward": 0.21875, + "rewards/format_reward_gqa": 0.96875, + "rewards/iou_glue_reward": 0.5753185749053955, + "step": 1624 + }, + { + "completion_length": 19.21875, + "epoch": 0.5178457616316124, + "grad_norm": 38.60752868652344, + "kl": 0.1787109375, + "learning_rate": 4.821542383683875e-07, + "loss": 0.0071, + "reward": 1.5983529090881348, + "reward_std": 0.05511268973350525, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5983529090881348, + "rewards/pad": 0.0, + "step": 1625 + }, + { + "completion_length": 121.5, + "epoch": 0.5181644359464627, + "grad_norm": 13.00921630859375, + "kl": 0.1025390625, + "learning_rate": 4.818355640535373e-07, + "loss": 0.0041, + "reward": 1.6340184211730957, + "reward_std": 0.10035441815853119, + "rewards/pad": 0.109375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5246434807777405, + "step": 1626 + }, + { + "completion_length": 123.125, + "epoch": 0.5184831102613129, + "grad_norm": 20.170867919921875, + "kl": 0.1259765625, + "learning_rate": 4.815168897386871e-07, + "loss": 0.005, + "reward": 1.5755319595336914, + "reward_std": 0.11915256828069687, + "rewards/pad": 0.09375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.48178181052207947, + "step": 1627 + }, + { + "completion_length": 97.046875, + "epoch": 0.5188017845761632, + "grad_norm": 26.876707077026367, + "kl": 0.140625, + "learning_rate": 4.811982154238368e-07, + "loss": 0.0056, + "reward": 1.606523036956787, + "reward_std": 0.12204252183437347, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.4971481263637543, + "rewards/pad": 0.125, + "step": 1628 + }, + { + "completion_length": 71.625, + "epoch": 0.5191204588910134, + "grad_norm": 27.261844635009766, + "kl": 0.203125, + "learning_rate": 4.808795411089866e-07, + "loss": 0.0081, + "reward": 1.6306061744689941, + "reward_std": 0.07760140299797058, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5056062340736389, + "rewards/pad": 0.125, + "step": 1629 + }, + { + "completion_length": 153.984375, + "epoch": 0.5194391332058637, + "grad_norm": 57.45818328857422, + "kl": 0.115234375, + "learning_rate": 4.805608667941364e-07, + "loss": 0.0046, + "reward": 1.380428433418274, + "reward_std": 0.1852729320526123, + "rewards/format_reward_tg": 0.96875, + "rewards/iou_timestamp_reward": 0.41167837381362915, + "rewards/pad": 0.0, + "step": 1630 + }, + { + "completion_length": 98.328125, + "epoch": 0.5197578075207139, + "grad_norm": 18.166400909423828, + "kl": 0.1455078125, + "learning_rate": 4.802421924792862e-07, + "loss": 0.0058, + "reward": 1.5990135669708252, + "reward_std": 0.10991062223911285, + "rewards/answer_reward": 0.109375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4896385669708252, + "step": 1631 + }, + { + "completion_length": 75.15625, + "epoch": 0.5200764818355641, + "grad_norm": 51.96231460571289, + "kl": 0.171875, + "learning_rate": 4.799235181644359e-07, + "loss": 0.0069, + "reward": 1.7394810914993286, + "reward_std": 0.06469561904668808, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6144810914993286, + "rewards/pad": 0.125, + "step": 1632 + }, + { + "completion_length": 120.40625, + "epoch": 0.5203951561504143, + "grad_norm": 32.042728424072266, + "kl": 0.11962890625, + "learning_rate": 4.796048438495857e-07, + "loss": 0.0048, + "reward": 1.4044157266616821, + "reward_std": 0.04880164563655853, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4044157564640045, + "step": 1633 + }, + { + "completion_length": 73.484375, + "epoch": 0.5207138304652645, + "grad_norm": 23.388429641723633, + "kl": 0.169921875, + "learning_rate": 4.792861695347355e-07, + "loss": 0.0068, + "reward": 1.5240478515625, + "reward_std": 0.09146015346050262, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5240478515625, + "step": 1634 + }, + { + "completion_length": 72.84375, + "epoch": 0.5210325047801148, + "grad_norm": 160.82843017578125, + "kl": 0.296875, + "learning_rate": 4.789674952198852e-07, + "loss": 0.0118, + "reward": 1.577231764793396, + "reward_std": 0.15222786366939545, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.48348167538642883, + "rewards/pad": 0.09375, + "step": 1635 + }, + { + "completion_length": 150.15625, + "epoch": 0.521351179094965, + "grad_norm": 13.156582832336426, + "kl": 0.07177734375, + "learning_rate": 4.78648820905035e-07, + "loss": 0.0029, + "reward": 1.542654037475586, + "reward_std": 0.03593272343277931, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4176541268825531, + "step": 1636 + }, + { + "completion_length": 200.171875, + "epoch": 0.5216698534098152, + "grad_norm": 14.329692840576172, + "kl": 0.0751953125, + "learning_rate": 4.783301465901848e-07, + "loss": 0.003, + "reward": 1.4973219633102417, + "reward_std": 0.04188467934727669, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4973219335079193, + "step": 1637 + }, + { + "completion_length": 74.78125, + "epoch": 0.5219885277246654, + "grad_norm": 33.7816047668457, + "kl": 0.2255859375, + "learning_rate": 4.780114722753345e-07, + "loss": 0.009, + "reward": 1.7144832611083984, + "reward_std": 0.06846581399440765, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5894832015037537, + "step": 1638 + }, + { + "completion_length": 120.890625, + "epoch": 0.5223072020395156, + "grad_norm": 25.760684967041016, + "kl": 0.1962890625, + "learning_rate": 4.776927979604843e-07, + "loss": 0.0078, + "reward": 1.4427735805511475, + "reward_std": 0.05109262466430664, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4427736699581146, + "rewards/pad": 0.0, + "step": 1639 + }, + { + "completion_length": 97.875, + "epoch": 0.5226258763543659, + "grad_norm": 30.939523696899414, + "kl": 0.24609375, + "learning_rate": 4.773741236456342e-07, + "loss": 0.0099, + "reward": 1.850117802619934, + "reward_std": 0.07939118146896362, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6001178622245789, + "rewards/pad": 0.25, + "step": 1640 + }, + { + "completion_length": 151.46875, + "epoch": 0.5229445506692161, + "grad_norm": 350.8528137207031, + "kl": 0.2119140625, + "learning_rate": 4.770554493307839e-07, + "loss": 0.0085, + "reward": 1.578183650970459, + "reward_std": 0.10141550749540329, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.46880874037742615, + "rewards/pad": 0.109375, + "step": 1641 + }, + { + "completion_length": 95.984375, + "epoch": 0.5232632249840663, + "grad_norm": 40.61670684814453, + "kl": 0.130859375, + "learning_rate": 4.7673677501593366e-07, + "loss": 0.0052, + "reward": 1.642643690109253, + "reward_std": 0.07146844267845154, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5176436305046082, + "rewards/pad": 0.125, + "step": 1642 + }, + { + "completion_length": 97.140625, + "epoch": 0.5235818992989165, + "grad_norm": 56.291839599609375, + "kl": 0.11767578125, + "learning_rate": 4.7641810070108347e-07, + "loss": 0.0047, + "reward": 1.4887385368347168, + "reward_std": 0.048854757100343704, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4887385070323944, + "rewards/pad": 0.0, + "step": 1643 + }, + { + "completion_length": 181.625, + "epoch": 0.5239005736137667, + "grad_norm": 6.92765998840332, + "kl": 0.11962890625, + "learning_rate": 4.760994263862332e-07, + "loss": 0.0048, + "reward": 1.4948139190673828, + "reward_std": 0.09755484759807587, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.3854389190673828, + "step": 1644 + }, + { + "completion_length": 69.96875, + "epoch": 0.524219247928617, + "grad_norm": 98.2040023803711, + "kl": 0.203125, + "learning_rate": 4.7578075207138303e-07, + "loss": 0.0081, + "reward": 1.4809551239013672, + "reward_std": 0.09209290891885757, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4809551537036896, + "rewards/pad": 0.0, + "step": 1645 + }, + { + "completion_length": 74.09375, + "epoch": 0.5245379222434672, + "grad_norm": 27.13360595703125, + "kl": 0.2333984375, + "learning_rate": 4.754620777565328e-07, + "loss": 0.0093, + "reward": 1.660394549369812, + "reward_std": 0.10714735835790634, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.535394549369812, + "step": 1646 + }, + { + "completion_length": 123.921875, + "epoch": 0.5248565965583174, + "grad_norm": 51.77692413330078, + "kl": 0.14453125, + "learning_rate": 4.751434034416826e-07, + "loss": 0.0058, + "reward": 1.386512279510498, + "reward_std": 0.07405687123537064, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3865121901035309, + "step": 1647 + }, + { + "completion_length": 122.96875, + "epoch": 0.5251752708731676, + "grad_norm": 57.70017623901367, + "kl": 0.10205078125, + "learning_rate": 4.7482472912683235e-07, + "loss": 0.0041, + "reward": 1.5424884557724, + "reward_std": 0.0725923627614975, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4174885153770447, + "rewards/pad": 0.125, + "step": 1648 + }, + { + "completion_length": 46.3125, + "epoch": 0.5254939451880178, + "grad_norm": 32.44889831542969, + "kl": 0.150390625, + "learning_rate": 4.7450605481198215e-07, + "loss": 0.006, + "reward": 1.6891388893127441, + "reward_std": 0.08782389760017395, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4391387701034546, + "step": 1649 + }, + { + "completion_length": 97.53125, + "epoch": 0.5258126195028681, + "grad_norm": 34.80462646484375, + "kl": 0.111328125, + "learning_rate": 4.741873804971319e-07, + "loss": 0.0044, + "reward": 1.542208194732666, + "reward_std": 0.06380252540111542, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.41720810532569885, + "step": 1650 + }, + { + "completion_length": 123.875, + "epoch": 0.5261312938177183, + "grad_norm": 41.51993942260742, + "kl": 0.1162109375, + "learning_rate": 4.738687061822817e-07, + "loss": 0.0047, + "reward": 1.470268964767456, + "reward_std": 0.050267286598682404, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4702690541744232, + "step": 1651 + }, + { + "completion_length": 123.453125, + "epoch": 0.5264499681325685, + "grad_norm": 55.49713897705078, + "kl": 0.12109375, + "learning_rate": 4.7355003186743147e-07, + "loss": 0.0048, + "reward": 1.4749317169189453, + "reward_std": 0.07620559632778168, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.47493165731430054, + "step": 1652 + }, + { + "completion_length": 72.265625, + "epoch": 0.5267686424474187, + "grad_norm": 83.41410827636719, + "kl": 0.1640625, + "learning_rate": 4.732313575525813e-07, + "loss": 0.0066, + "reward": 1.6315970420837402, + "reward_std": 0.1632934808731079, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6003470420837402, + "rewards/pad": 0.03125, + "step": 1653 + }, + { + "completion_length": 100.109375, + "epoch": 0.5270873167622689, + "grad_norm": 29.058053970336914, + "kl": 0.15625, + "learning_rate": 4.7291268323773103e-07, + "loss": 0.0062, + "reward": 1.5368297100067139, + "reward_std": 0.1058320552110672, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.41182976961135864, + "step": 1654 + }, + { + "completion_length": 45.09375, + "epoch": 0.5274059910771192, + "grad_norm": 31.651737213134766, + "kl": 0.1591796875, + "learning_rate": 4.7259400892288084e-07, + "loss": 0.0064, + "reward": 1.491950511932373, + "reward_std": 0.06471245735883713, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4919503927230835, + "rewards/pad": 0.0, + "step": 1655 + }, + { + "completion_length": 70.6875, + "epoch": 0.5277246653919694, + "grad_norm": 48.495574951171875, + "kl": 0.1533203125, + "learning_rate": 4.722753346080306e-07, + "loss": 0.0061, + "reward": 1.7000138759613037, + "reward_std": 0.11548508703708649, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5750138163566589, + "rewards/pad": 0.125, + "step": 1656 + }, + { + "completion_length": 96.0625, + "epoch": 0.5280433397068196, + "grad_norm": 42.84994888305664, + "kl": 0.310546875, + "learning_rate": 4.719566602931804e-07, + "loss": 0.0124, + "reward": 1.5187783241271973, + "reward_std": 0.04149685800075531, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5187783241271973, + "step": 1657 + }, + { + "completion_length": 97.265625, + "epoch": 0.5283620140216698, + "grad_norm": 28.610042572021484, + "kl": 0.1630859375, + "learning_rate": 4.716379859783301e-07, + "loss": 0.0065, + "reward": 1.5452983379364014, + "reward_std": 0.07879804074764252, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4202982485294342, + "step": 1658 + }, + { + "completion_length": 122.921875, + "epoch": 0.52868068833652, + "grad_norm": 51.03239440917969, + "kl": 0.1171875, + "learning_rate": 4.713193116634799e-07, + "loss": 0.0047, + "reward": 1.5429987907409668, + "reward_std": 0.1555194854736328, + "rewards/format_reward_tg": 0.96875, + "rewards/iou_timestamp_reward": 0.44924867153167725, + "rewards/pad": 0.125, + "step": 1659 + }, + { + "completion_length": 98.25, + "epoch": 0.5289993626513703, + "grad_norm": 53.16851043701172, + "kl": 0.126953125, + "learning_rate": 4.7100063734862966e-07, + "loss": 0.0051, + "reward": 1.633514642715454, + "reward_std": 0.07731299102306366, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5085147619247437, + "step": 1660 + }, + { + "completion_length": 72.828125, + "epoch": 0.5293180369662205, + "grad_norm": 19.036083221435547, + "kl": 0.12109375, + "learning_rate": 4.706819630337794e-07, + "loss": 0.0048, + "reward": 1.5364749431610107, + "reward_std": 0.0853765681385994, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3958500325679779, + "rewards/pad": 0.140625, + "step": 1661 + }, + { + "completion_length": 94.09375, + "epoch": 0.5296367112810707, + "grad_norm": 31.02826499938965, + "kl": 0.125, + "learning_rate": 4.703632887189292e-07, + "loss": 0.005, + "reward": 1.4983229637145996, + "reward_std": 0.06386838108301163, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4983229339122772, + "rewards/pad": 0.0, + "step": 1662 + }, + { + "completion_length": 118.890625, + "epoch": 0.5299553855959209, + "grad_norm": 30.89619255065918, + "kl": 0.10693359375, + "learning_rate": 4.70044614404079e-07, + "loss": 0.0043, + "reward": 1.4212117195129395, + "reward_std": 0.04871714487671852, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.42121171951293945, + "step": 1663 + }, + { + "completion_length": 96.296875, + "epoch": 0.5302740599107711, + "grad_norm": 25.728801727294922, + "kl": 0.1259765625, + "learning_rate": 4.697259400892288e-07, + "loss": 0.005, + "reward": 1.4896304607391357, + "reward_std": 0.08095531165599823, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.48963046073913574, + "rewards/pad": 0.0, + "step": 1664 + }, + { + "completion_length": 125.40625, + "epoch": 0.5305927342256214, + "grad_norm": 18.65762710571289, + "kl": 0.07080078125, + "learning_rate": 4.6940726577437853e-07, + "loss": 0.0028, + "reward": 1.63633394241333, + "reward_std": 0.07638265937566757, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5113338828086853, + "step": 1665 + }, + { + "completion_length": 122.71875, + "epoch": 0.5309114085404716, + "grad_norm": 13.211691856384277, + "kl": 0.091796875, + "learning_rate": 4.6908859145952834e-07, + "loss": 0.0037, + "reward": 1.618811845779419, + "reward_std": 0.05741771683096886, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4938117265701294, + "step": 1666 + }, + { + "completion_length": 95.46875, + "epoch": 0.5312300828553218, + "grad_norm": 32.33008575439453, + "kl": 0.2041015625, + "learning_rate": 4.687699171446781e-07, + "loss": 0.0082, + "reward": 1.3568158149719238, + "reward_std": 0.03793134540319443, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.35681572556495667, + "rewards/pad": 0.0, + "step": 1667 + }, + { + "completion_length": 96.359375, + "epoch": 0.5315487571701721, + "grad_norm": 26.959739685058594, + "kl": 0.11083984375, + "learning_rate": 4.684512428298279e-07, + "loss": 0.0044, + "reward": 1.5582318305969238, + "reward_std": 0.08168759942054749, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5582318902015686, + "rewards/pad": 0.0, + "step": 1668 + }, + { + "completion_length": 70.75, + "epoch": 0.5318674314850224, + "grad_norm": 106.21484375, + "kl": 0.1513671875, + "learning_rate": 4.6813256851497766e-07, + "loss": 0.006, + "reward": 1.5603129863739014, + "reward_std": 0.08481752872467041, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.45093798637390137, + "rewards/pad": 0.109375, + "step": 1669 + }, + { + "completion_length": 97.265625, + "epoch": 0.5321861057998726, + "grad_norm": 31.24065589904785, + "kl": 0.12060546875, + "learning_rate": 4.6781389420012746e-07, + "loss": 0.0048, + "reward": 1.829498529434204, + "reward_std": 0.07612404227256775, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5794985890388489, + "step": 1670 + }, + { + "completion_length": 146.171875, + "epoch": 0.5325047801147228, + "grad_norm": 22.557510375976562, + "kl": 0.08203125, + "learning_rate": 4.674952198852772e-07, + "loss": 0.0033, + "reward": 1.5544493198394775, + "reward_std": 0.0435752347111702, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5544492602348328, + "step": 1671 + }, + { + "completion_length": 96.65625, + "epoch": 0.532823454429573, + "grad_norm": 16.112016677856445, + "kl": 0.111328125, + "learning_rate": 4.67176545570427e-07, + "loss": 0.0044, + "reward": 1.5003379583358765, + "reward_std": 0.0549815371632576, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5003380179405212, + "step": 1672 + }, + { + "completion_length": 70.046875, + "epoch": 0.5331421287444232, + "grad_norm": 38.697509765625, + "kl": 0.2333984375, + "learning_rate": 4.668578712555768e-07, + "loss": 0.0093, + "reward": 1.4883949756622314, + "reward_std": 0.12367160618305206, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.48839494585990906, + "rewards/pad": 0.0, + "step": 1673 + }, + { + "completion_length": 73.140625, + "epoch": 0.5334608030592735, + "grad_norm": 45.91947555541992, + "kl": 0.1513671875, + "learning_rate": 4.665391969407266e-07, + "loss": 0.0061, + "reward": 1.7846755981445312, + "reward_std": 0.1634289026260376, + "rewards/answer_reward": 0.40625, + "rewards/format_reward_gqa": 0.984375, + "rewards/iou_glue_reward": 0.394050657749176, + "step": 1674 + }, + { + "completion_length": 97.578125, + "epoch": 0.5337794773741237, + "grad_norm": 70.14591979980469, + "kl": 0.15234375, + "learning_rate": 4.6622052262587634e-07, + "loss": 0.0061, + "reward": 1.4238718748092651, + "reward_std": 0.0894370749592781, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.42387187480926514, + "rewards/pad": 0.0, + "step": 1675 + }, + { + "completion_length": 121.34375, + "epoch": 0.5340981516889739, + "grad_norm": 9.284736633300781, + "kl": 0.1103515625, + "learning_rate": 4.6590184831102615e-07, + "loss": 0.0044, + "reward": 1.562800407409668, + "reward_std": 0.05218476057052612, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5628004670143127, + "rewards/pad": 0.0, + "step": 1676 + }, + { + "completion_length": 124.09375, + "epoch": 0.5344168260038241, + "grad_norm": 13.18323040008545, + "kl": 0.154296875, + "learning_rate": 4.6558317399617585e-07, + "loss": 0.0062, + "reward": 1.4828343391418457, + "reward_std": 0.1092568188905716, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.46720945835113525, + "rewards/pad": 0.015625, + "step": 1677 + }, + { + "completion_length": 122.71875, + "epoch": 0.5347355003186743, + "grad_norm": 56.83836364746094, + "kl": 0.11572265625, + "learning_rate": 4.6526449968132566e-07, + "loss": 0.0046, + "reward": 1.644931674003601, + "reward_std": 0.11709925532341003, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 0.984375, + "rewards/iou_glue_reward": 0.5355565547943115, + "step": 1678 + }, + { + "completion_length": 173.828125, + "epoch": 0.5350541746335246, + "grad_norm": 6.490396976470947, + "kl": 0.0771484375, + "learning_rate": 4.649458253664754e-07, + "loss": 0.0031, + "reward": 1.646160364151001, + "reward_std": 0.03755709156394005, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.6461603045463562, + "step": 1679 + }, + { + "completion_length": 96.375, + "epoch": 0.5353728489483748, + "grad_norm": 31.79176902770996, + "kl": 0.154296875, + "learning_rate": 4.646271510516252e-07, + "loss": 0.0062, + "reward": 1.514885425567627, + "reward_std": 0.05413583666086197, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.514885425567627, + "rewards/pad": 0.0, + "step": 1680 + }, + { + "completion_length": 122.1875, + "epoch": 0.535691523263225, + "grad_norm": 16.86212158203125, + "kl": 0.15625, + "learning_rate": 4.6430847673677497e-07, + "loss": 0.0063, + "reward": 1.5246326923370361, + "reward_std": 0.09441748261451721, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5246326923370361, + "rewards/pad": 0.0, + "step": 1681 + }, + { + "completion_length": 68.9375, + "epoch": 0.5360101975780752, + "grad_norm": 40.876766204833984, + "kl": 0.1806640625, + "learning_rate": 4.639898024219248e-07, + "loss": 0.0072, + "reward": 1.5343561172485352, + "reward_std": 0.06397495418787003, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5343560576438904, + "rewards/pad": 0.0, + "step": 1682 + }, + { + "completion_length": 97.75, + "epoch": 0.5363288718929254, + "grad_norm": 28.436660766601562, + "kl": 0.11279296875, + "learning_rate": 4.6367112810707453e-07, + "loss": 0.0045, + "reward": 1.4423682689666748, + "reward_std": 0.05234856903553009, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.44236820936203003, + "step": 1683 + }, + { + "completion_length": 20.625, + "epoch": 0.5366475462077757, + "grad_norm": 33.24636459350586, + "kl": 0.2041015625, + "learning_rate": 4.6335245379222434e-07, + "loss": 0.0082, + "reward": 1.596472978591919, + "reward_std": 0.09440158307552338, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4714728891849518, + "rewards/pad": 0.125, + "step": 1684 + }, + { + "completion_length": 148.65625, + "epoch": 0.5369662205226259, + "grad_norm": 37.23117446899414, + "kl": 0.2099609375, + "learning_rate": 4.630337794773741e-07, + "loss": 0.0084, + "reward": 1.5583281517028809, + "reward_std": 0.07235017418861389, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.43332818150520325, + "step": 1685 + }, + { + "completion_length": 71.5, + "epoch": 0.5372848948374761, + "grad_norm": 33.29726791381836, + "kl": 0.1376953125, + "learning_rate": 4.627151051625239e-07, + "loss": 0.0055, + "reward": 1.618628740310669, + "reward_std": 0.09640872478485107, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4936287999153137, + "rewards/pad": 0.125, + "step": 1686 + }, + { + "completion_length": 121.78125, + "epoch": 0.5376035691523263, + "grad_norm": 9.005082130432129, + "kl": 0.1181640625, + "learning_rate": 4.6239643084767365e-07, + "loss": 0.0047, + "reward": 1.4646806716918945, + "reward_std": 0.052319884300231934, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.464680552482605, + "step": 1687 + }, + { + "completion_length": 71.21875, + "epoch": 0.5379222434671765, + "grad_norm": 26.622392654418945, + "kl": 0.1728515625, + "learning_rate": 4.6207775653282346e-07, + "loss": 0.0069, + "reward": 1.6820809841156006, + "reward_std": 0.09217497706413269, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.6820809841156006, + "step": 1688 + }, + { + "completion_length": 123.609375, + "epoch": 0.5382409177820268, + "grad_norm": 97.50489044189453, + "kl": 0.185546875, + "learning_rate": 4.617590822179732e-07, + "loss": 0.0074, + "reward": 1.5759509801864624, + "reward_std": 0.12089285254478455, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4509509205818176, + "step": 1689 + }, + { + "completion_length": 94.3125, + "epoch": 0.538559592096877, + "grad_norm": 39.29410934448242, + "kl": 0.5078125, + "learning_rate": 4.61440407903123e-07, + "loss": 0.0204, + "reward": 1.592628002166748, + "reward_std": 0.08428885787725449, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5926278829574585, + "rewards/pad": 0.0, + "step": 1690 + }, + { + "completion_length": 94.984375, + "epoch": 0.5388782664117272, + "grad_norm": 27.28589630126953, + "kl": 0.19140625, + "learning_rate": 4.611217335882728e-07, + "loss": 0.0077, + "reward": 1.4897959232330322, + "reward_std": 0.061947956681251526, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.48979589343070984, + "rewards/pad": 0.0, + "step": 1691 + }, + { + "completion_length": 69.59375, + "epoch": 0.5391969407265774, + "grad_norm": 15.837658882141113, + "kl": 0.314453125, + "learning_rate": 4.608030592734226e-07, + "loss": 0.0126, + "reward": 1.4092113971710205, + "reward_std": 0.045040152966976166, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.28421148657798767, + "step": 1692 + }, + { + "completion_length": 69.234375, + "epoch": 0.5395156150414276, + "grad_norm": 26.907012939453125, + "kl": 0.162109375, + "learning_rate": 4.6048438495857234e-07, + "loss": 0.0065, + "reward": 1.5673654079437256, + "reward_std": 0.09178591519594193, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5673654079437256, + "rewards/pad": 0.0, + "step": 1693 + }, + { + "completion_length": 97.015625, + "epoch": 0.5398342893562779, + "grad_norm": 33.33298873901367, + "kl": 0.142578125, + "learning_rate": 4.601657106437221e-07, + "loss": 0.0057, + "reward": 1.7240219116210938, + "reward_std": 0.10178350657224655, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5990219712257385, + "rewards/pad": 0.125, + "step": 1694 + }, + { + "completion_length": 147.5625, + "epoch": 0.5401529636711281, + "grad_norm": 15.036523818969727, + "kl": 0.1083984375, + "learning_rate": 4.598470363288719e-07, + "loss": 0.0043, + "reward": 1.528503179550171, + "reward_std": 0.05876230448484421, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5285031795501709, + "rewards/pad": 0.0, + "step": 1695 + }, + { + "completion_length": 96.390625, + "epoch": 0.5404716379859783, + "grad_norm": 20.0390682220459, + "kl": 0.173828125, + "learning_rate": 4.5952836201402165e-07, + "loss": 0.007, + "reward": 1.6498191356658936, + "reward_std": 0.058616429567337036, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3998190462589264, + "rewards/pad": 0.25, + "step": 1696 + }, + { + "completion_length": 97.6875, + "epoch": 0.5407903123008285, + "grad_norm": 100.7648696899414, + "kl": 0.158203125, + "learning_rate": 4.592096876991714e-07, + "loss": 0.0063, + "reward": 1.6984670162200928, + "reward_std": 0.10114402323961258, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5265920162200928, + "rewards/pad": 0.171875, + "step": 1697 + }, + { + "completion_length": 93.28125, + "epoch": 0.5411089866156787, + "grad_norm": 18.635839462280273, + "kl": 0.1865234375, + "learning_rate": 4.5889101338432116e-07, + "loss": 0.0074, + "reward": 1.5091419219970703, + "reward_std": 0.06434226781129837, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5091418623924255, + "step": 1698 + }, + { + "completion_length": 95.71875, + "epoch": 0.541427660930529, + "grad_norm": 148.67578125, + "kl": 0.1328125, + "learning_rate": 4.5857233906947097e-07, + "loss": 0.0053, + "reward": 1.6802880764007568, + "reward_std": 0.06698621809482574, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.6802880764007568, + "step": 1699 + }, + { + "completion_length": 124.125, + "epoch": 0.5417463352453792, + "grad_norm": 8.406394958496094, + "kl": 0.091796875, + "learning_rate": 4.582536647546207e-07, + "loss": 0.0037, + "reward": 1.604690670967102, + "reward_std": 0.06378532201051712, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.47969070076942444, + "step": 1700 + }, + { + "completion_length": 46.109375, + "epoch": 0.5420650095602294, + "grad_norm": 23.811153411865234, + "kl": 0.138671875, + "learning_rate": 4.5793499043977053e-07, + "loss": 0.0056, + "reward": 1.6016175746917725, + "reward_std": 0.05782673507928848, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4766176640987396, + "rewards/pad": 0.125, + "step": 1701 + }, + { + "completion_length": 47.765625, + "epoch": 0.5423836838750796, + "grad_norm": 20.4882755279541, + "kl": 0.1748046875, + "learning_rate": 4.576163161249203e-07, + "loss": 0.007, + "reward": 2.0038671493530273, + "reward_std": 0.07229779660701752, + "rewards/answer_reward": 0.375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.6288673281669617, + "step": 1702 + }, + { + "completion_length": 98.109375, + "epoch": 0.5427023581899298, + "grad_norm": 8.164681434631348, + "kl": 0.1396484375, + "learning_rate": 4.572976418100701e-07, + "loss": 0.0056, + "reward": 1.90226411819458, + "reward_std": 0.06600087881088257, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.6522639989852905, + "step": 1703 + }, + { + "completion_length": 122.890625, + "epoch": 0.5430210325047801, + "grad_norm": 10.439600944519043, + "kl": 0.1513671875, + "learning_rate": 4.5697896749521984e-07, + "loss": 0.006, + "reward": 1.6205843687057495, + "reward_std": 0.07790669053792953, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6205843687057495, + "rewards/pad": 0.0, + "step": 1704 + }, + { + "completion_length": 97.984375, + "epoch": 0.5433397068196303, + "grad_norm": 16.746431350708008, + "kl": 0.12890625, + "learning_rate": 4.5666029318036965e-07, + "loss": 0.0052, + "reward": 1.438126802444458, + "reward_std": 0.037393443286418915, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4381268322467804, + "rewards/pad": 0.0, + "step": 1705 + }, + { + "completion_length": 174.625, + "epoch": 0.5436583811344805, + "grad_norm": 119.52044677734375, + "kl": 0.1025390625, + "learning_rate": 4.563416188655194e-07, + "loss": 0.0041, + "reward": 1.4924670457839966, + "reward_std": 0.04576924815773964, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4924669861793518, + "rewards/pad": 0.0, + "step": 1706 + }, + { + "completion_length": 120.671875, + "epoch": 0.5439770554493308, + "grad_norm": 27.163732528686523, + "kl": 0.11474609375, + "learning_rate": 4.560229445506692e-07, + "loss": 0.0046, + "reward": 1.5915979146957397, + "reward_std": 0.0663021057844162, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.48222291469573975, + "rewards/pad": 0.109375, + "step": 1707 + }, + { + "completion_length": 121.15625, + "epoch": 0.5442957297641811, + "grad_norm": 19.470294952392578, + "kl": 0.2158203125, + "learning_rate": 4.5570427023581896e-07, + "loss": 0.0086, + "reward": 1.5954852104187012, + "reward_std": 0.1103629320859909, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5954852104187012, + "rewards/pad": 0.0, + "step": 1708 + }, + { + "completion_length": 45.359375, + "epoch": 0.5446144040790313, + "grad_norm": 35.75319290161133, + "kl": 0.1689453125, + "learning_rate": 4.5538559592096877e-07, + "loss": 0.0068, + "reward": 1.6164058446884155, + "reward_std": 0.1098790392279625, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5226558446884155, + "rewards/pad": 0.09375, + "step": 1709 + }, + { + "completion_length": 122.734375, + "epoch": 0.5449330783938815, + "grad_norm": 77.64950561523438, + "kl": 0.1826171875, + "learning_rate": 4.550669216061185e-07, + "loss": 0.0073, + "reward": 1.546919584274292, + "reward_std": 0.0800800621509552, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.421919584274292, + "rewards/pad": 0.125, + "step": 1710 + }, + { + "completion_length": 98.78125, + "epoch": 0.5452517527087317, + "grad_norm": 16.44868278503418, + "kl": 0.154296875, + "learning_rate": 4.5474824729126833e-07, + "loss": 0.0062, + "reward": 1.552331566810608, + "reward_std": 0.040778547525405884, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4273315668106079, + "step": 1711 + }, + { + "completion_length": 71.4375, + "epoch": 0.5455704270235819, + "grad_norm": 97.36717987060547, + "kl": 0.173828125, + "learning_rate": 4.544295729764181e-07, + "loss": 0.007, + "reward": 1.6137738227844238, + "reward_std": 0.06848740577697754, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.48877376317977905, + "rewards/pad": 0.125, + "step": 1712 + }, + { + "completion_length": 70.84375, + "epoch": 0.5458891013384322, + "grad_norm": 26.707841873168945, + "kl": 0.1943359375, + "learning_rate": 4.541108986615679e-07, + "loss": 0.0078, + "reward": 1.5837278366088867, + "reward_std": 0.06415103375911713, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4587279260158539, + "rewards/pad": 0.125, + "step": 1713 + }, + { + "completion_length": 95.40625, + "epoch": 0.5462077756532824, + "grad_norm": 81.21223449707031, + "kl": 0.2265625, + "learning_rate": 4.5379222434671765e-07, + "loss": 0.009, + "reward": 1.4255642890930176, + "reward_std": 0.05765555426478386, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.425564169883728, + "rewards/pad": 0.0, + "step": 1714 + }, + { + "completion_length": 96.625, + "epoch": 0.5465264499681326, + "grad_norm": 17.838123321533203, + "kl": 0.12158203125, + "learning_rate": 4.5347355003186745e-07, + "loss": 0.0049, + "reward": 1.4614139795303345, + "reward_std": 0.041417405009269714, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4614139795303345, + "step": 1715 + }, + { + "completion_length": 98.140625, + "epoch": 0.5468451242829828, + "grad_norm": 16.121427536010742, + "kl": 0.25390625, + "learning_rate": 4.531548757170172e-07, + "loss": 0.0101, + "reward": 1.5973544120788574, + "reward_std": 0.08118489384651184, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.45672938227653503, + "rewards/pad": 0.140625, + "step": 1716 + }, + { + "completion_length": 69.5625, + "epoch": 0.547163798597833, + "grad_norm": 114.39732360839844, + "kl": 0.14453125, + "learning_rate": 4.5283620140216696e-07, + "loss": 0.0058, + "reward": 1.7027480602264404, + "reward_std": 0.07084536552429199, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.7027480006217957, + "step": 1717 + }, + { + "completion_length": 71.390625, + "epoch": 0.5474824729126833, + "grad_norm": 18.393924713134766, + "kl": 0.2265625, + "learning_rate": 4.525175270873167e-07, + "loss": 0.0091, + "reward": 1.7171638011932373, + "reward_std": 0.09167201817035675, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4671638309955597, + "rewards/pad": 0.25, + "step": 1718 + }, + { + "completion_length": 122.40625, + "epoch": 0.5478011472275335, + "grad_norm": 39.043643951416016, + "kl": 0.1201171875, + "learning_rate": 4.521988527724665e-07, + "loss": 0.0048, + "reward": 1.5978953838348389, + "reward_std": 0.069697305560112, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.47289541363716125, + "step": 1719 + }, + { + "completion_length": 147.8125, + "epoch": 0.5481198215423837, + "grad_norm": 26.979312896728516, + "kl": 0.138671875, + "learning_rate": 4.518801784576163e-07, + "loss": 0.0055, + "reward": 1.5596652030944824, + "reward_std": 0.0610220730304718, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5596652626991272, + "rewards/pad": 0.0, + "step": 1720 + }, + { + "completion_length": 97.0625, + "epoch": 0.5484384958572339, + "grad_norm": 35.1412353515625, + "kl": 0.197265625, + "learning_rate": 4.515615041427661e-07, + "loss": 0.0079, + "reward": 1.400559902191162, + "reward_std": 0.0762283131480217, + "rewards/answer_reward": 0.0, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4005598723888397, + "step": 1721 + }, + { + "completion_length": 97.09375, + "epoch": 0.5487571701720841, + "grad_norm": 40.99324417114258, + "kl": 0.125, + "learning_rate": 4.5124282982791584e-07, + "loss": 0.005, + "reward": 1.4977219104766846, + "reward_std": 0.1251518428325653, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4195968806743622, + "rewards/pad": 0.078125, + "step": 1722 + }, + { + "completion_length": 95.0, + "epoch": 0.5490758444869344, + "grad_norm": 27.61468505859375, + "kl": 0.134765625, + "learning_rate": 4.509241555130656e-07, + "loss": 0.0054, + "reward": 1.4169597625732422, + "reward_std": 0.16255837678909302, + "rewards/pad": 0.0625, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.35445982217788696, + "step": 1723 + }, + { + "completion_length": 95.25, + "epoch": 0.5493945188017846, + "grad_norm": 51.87438201904297, + "kl": 0.16015625, + "learning_rate": 4.506054811982154e-07, + "loss": 0.0064, + "reward": 1.3904025554656982, + "reward_std": 0.06188996881246567, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3904026448726654, + "step": 1724 + }, + { + "completion_length": 70.859375, + "epoch": 0.5497131931166348, + "grad_norm": 47.93151092529297, + "kl": 0.1396484375, + "learning_rate": 4.5028680688336515e-07, + "loss": 0.0056, + "reward": 1.4469374418258667, + "reward_std": 0.05983661487698555, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4469375014305115, + "rewards/pad": 0.0, + "step": 1725 + }, + { + "completion_length": 70.828125, + "epoch": 0.550031867431485, + "grad_norm": 55.93996047973633, + "kl": 0.2119140625, + "learning_rate": 4.4996813256851496e-07, + "loss": 0.0085, + "reward": 1.595890998840332, + "reward_std": 0.11610599607229233, + "rewards/answer_reward": 0.0625, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5333911180496216, + "step": 1726 + }, + { + "completion_length": 122.890625, + "epoch": 0.5503505417463352, + "grad_norm": 29.33034896850586, + "kl": 0.13671875, + "learning_rate": 4.496494582536647e-07, + "loss": 0.0055, + "reward": 1.574709177017212, + "reward_std": 0.0751732587814331, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5747092366218567, + "step": 1727 + }, + { + "completion_length": 72.0, + "epoch": 0.5506692160611855, + "grad_norm": 45.549583435058594, + "kl": 0.1357421875, + "learning_rate": 4.493307839388145e-07, + "loss": 0.0054, + "reward": 1.8912920951843262, + "reward_std": 0.10724160820245743, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5475420355796814, + "rewards/pad": 0.34375, + "step": 1728 + }, + { + "completion_length": 122.859375, + "epoch": 0.5509878903760357, + "grad_norm": 122.186767578125, + "kl": 0.0888671875, + "learning_rate": 4.490121096239643e-07, + "loss": 0.0036, + "reward": 1.504943609237671, + "reward_std": 0.09938017278909683, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5049434900283813, + "rewards/pad": 0.0, + "step": 1729 + }, + { + "completion_length": 44.203125, + "epoch": 0.5513065646908859, + "grad_norm": 33.6113166809082, + "kl": 0.17578125, + "learning_rate": 4.486934353091141e-07, + "loss": 0.007, + "reward": 1.710925579071045, + "reward_std": 0.08198478817939758, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.7109256982803345, + "rewards/pad": 0.0, + "step": 1730 + }, + { + "completion_length": 43.546875, + "epoch": 0.5516252390057361, + "grad_norm": 20.308115005493164, + "kl": 0.484375, + "learning_rate": 4.4837476099426384e-07, + "loss": 0.0194, + "reward": 1.6070808172225952, + "reward_std": 0.09324438869953156, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6070806980133057, + "rewards/pad": 0.0, + "step": 1731 + }, + { + "completion_length": 119.421875, + "epoch": 0.5519439133205863, + "grad_norm": 461.97021484375, + "kl": 0.134765625, + "learning_rate": 4.4805608667941364e-07, + "loss": 0.0054, + "reward": 1.552412509918213, + "reward_std": 0.05798398330807686, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5524123907089233, + "rewards/pad": 0.0, + "step": 1732 + }, + { + "completion_length": 96.515625, + "epoch": 0.5522625876354366, + "grad_norm": 27.874906539916992, + "kl": 0.16015625, + "learning_rate": 4.477374123645634e-07, + "loss": 0.0064, + "reward": 1.422250747680664, + "reward_std": 0.050203837454319, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.42225080728530884, + "rewards/pad": 0.0, + "step": 1733 + }, + { + "completion_length": 118.59375, + "epoch": 0.5525812619502868, + "grad_norm": 25.63271713256836, + "kl": 0.12890625, + "learning_rate": 4.474187380497132e-07, + "loss": 0.0051, + "reward": 1.4211066961288452, + "reward_std": 0.052193887531757355, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.42110663652420044, + "rewards/pad": 0.0, + "step": 1734 + }, + { + "completion_length": 44.1875, + "epoch": 0.552899936265137, + "grad_norm": 34.65806579589844, + "kl": 0.166015625, + "learning_rate": 4.4710006373486296e-07, + "loss": 0.0067, + "reward": 1.8596563339233398, + "reward_std": 0.08217694610357285, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.6096563339233398, + "step": 1735 + }, + { + "completion_length": 94.078125, + "epoch": 0.5532186105799872, + "grad_norm": 33.38858413696289, + "kl": 0.1083984375, + "learning_rate": 4.467813894200127e-07, + "loss": 0.0043, + "reward": 1.6893126964569092, + "reward_std": 0.04501698166131973, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.439312607049942, + "step": 1736 + }, + { + "completion_length": 72.140625, + "epoch": 0.5535372848948374, + "grad_norm": 38.28264236450195, + "kl": 0.1455078125, + "learning_rate": 4.4646271510516247e-07, + "loss": 0.0058, + "reward": 1.7901391983032227, + "reward_std": 0.048542171716690063, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5401391386985779, + "rewards/pad": 0.25, + "step": 1737 + }, + { + "completion_length": 70.671875, + "epoch": 0.5538559592096877, + "grad_norm": 49.95315170288086, + "kl": 0.17578125, + "learning_rate": 4.4614404079031227e-07, + "loss": 0.007, + "reward": 1.5110564231872559, + "reward_std": 0.08883722126483917, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5110564231872559, + "step": 1738 + }, + { + "completion_length": 95.1875, + "epoch": 0.5541746335245379, + "grad_norm": 79.57206726074219, + "kl": 0.11572265625, + "learning_rate": 4.4582536647546203e-07, + "loss": 0.0046, + "reward": 1.468375325202942, + "reward_std": 0.03867108374834061, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4683753252029419, + "step": 1739 + }, + { + "completion_length": 124.90625, + "epoch": 0.5544933078393881, + "grad_norm": 50.3587532043457, + "kl": 0.1083984375, + "learning_rate": 4.4550669216061183e-07, + "loss": 0.0043, + "reward": 1.4035348892211914, + "reward_std": 0.03411925584077835, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.403534859418869, + "rewards/pad": 0.0, + "step": 1740 + }, + { + "completion_length": 43.03125, + "epoch": 0.5548119821542383, + "grad_norm": 42.346439361572266, + "kl": 0.240234375, + "learning_rate": 4.451880178457616e-07, + "loss": 0.0096, + "reward": 1.7896267175674438, + "reward_std": 0.057393234223127365, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.6646266579627991, + "step": 1741 + }, + { + "completion_length": 121.8125, + "epoch": 0.5551306564690885, + "grad_norm": 18.29889678955078, + "kl": 0.197265625, + "learning_rate": 4.448693435309114e-07, + "loss": 0.0079, + "reward": 1.6041014194488525, + "reward_std": 0.15853551030158997, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.49472635984420776, + "step": 1742 + }, + { + "completion_length": 20.484375, + "epoch": 0.5554493307839388, + "grad_norm": 34.001243591308594, + "kl": 0.25, + "learning_rate": 4.4455066921606115e-07, + "loss": 0.01, + "reward": 1.8880293369293213, + "reward_std": 0.08389134705066681, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6380293965339661, + "rewards/pad": 0.25, + "step": 1743 + }, + { + "completion_length": 70.875, + "epoch": 0.555768005098789, + "grad_norm": 25.211380004882812, + "kl": 0.13671875, + "learning_rate": 4.4423199490121096e-07, + "loss": 0.0055, + "reward": 1.6173521280288696, + "reward_std": 0.11482757329940796, + "rewards/pad": 0.109375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5079771876335144, + "step": 1744 + }, + { + "completion_length": 43.640625, + "epoch": 0.5560866794136392, + "grad_norm": 33.18324661254883, + "kl": 0.1474609375, + "learning_rate": 4.439133205863607e-07, + "loss": 0.0059, + "reward": 1.6932084560394287, + "reward_std": 0.06315641105175018, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6932085156440735, + "rewards/pad": 0.0, + "step": 1745 + }, + { + "completion_length": 70.59375, + "epoch": 0.5564053537284895, + "grad_norm": 23.28170394897461, + "kl": 0.1064453125, + "learning_rate": 4.435946462715105e-07, + "loss": 0.0043, + "reward": 1.636433720588684, + "reward_std": 0.06571288406848907, + "rewards/answer_reward": 0.0, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.6364337205886841, + "step": 1746 + }, + { + "completion_length": 119.765625, + "epoch": 0.5567240280433398, + "grad_norm": 30.740108489990234, + "kl": 0.240234375, + "learning_rate": 4.4327597195666027e-07, + "loss": 0.0096, + "reward": 1.7088111639022827, + "reward_std": 0.10138913244009018, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5838110446929932, + "step": 1747 + }, + { + "completion_length": 73.3125, + "epoch": 0.55704270235819, + "grad_norm": 92.40509796142578, + "kl": 0.189453125, + "learning_rate": 4.429572976418101e-07, + "loss": 0.0076, + "reward": 1.7267744541168213, + "reward_std": 0.2050127536058426, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.5080245137214661, + "rewards/pad": 0.234375, + "step": 1748 + }, + { + "completion_length": 70.28125, + "epoch": 0.5573613766730402, + "grad_norm": 48.15990447998047, + "kl": 0.1494140625, + "learning_rate": 4.4263862332695983e-07, + "loss": 0.006, + "reward": 1.694014072418213, + "reward_std": 0.09263955056667328, + "rewards/pad": 0.109375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5846391320228577, + "step": 1749 + }, + { + "completion_length": 45.21875, + "epoch": 0.5576800509878904, + "grad_norm": 46.90249252319336, + "kl": 0.35546875, + "learning_rate": 4.4231994901210964e-07, + "loss": 0.0142, + "reward": 1.600165605545044, + "reward_std": 0.15194952487945557, + "rewards/answer_reward": 0.171875, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4282906651496887, + "step": 1750 + }, + { + "completion_length": 68.71875, + "epoch": 0.5579987253027406, + "grad_norm": 16.717771530151367, + "kl": 0.1533203125, + "learning_rate": 4.420012746972594e-07, + "loss": 0.0061, + "reward": 1.71664297580719, + "reward_std": 0.07848171889781952, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5916429758071899, + "rewards/pad": 0.125, + "step": 1751 + }, + { + "completion_length": 122.171875, + "epoch": 0.5583173996175909, + "grad_norm": 96.13909149169922, + "kl": 0.1865234375, + "learning_rate": 4.416826003824092e-07, + "loss": 0.0075, + "reward": 1.4779011011123657, + "reward_std": 0.159339040517807, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.3685261309146881, + "step": 1752 + }, + { + "completion_length": 173.46875, + "epoch": 0.5586360739324411, + "grad_norm": 43.84035110473633, + "kl": 0.09228515625, + "learning_rate": 4.4136392606755895e-07, + "loss": 0.0037, + "reward": 1.5944308042526245, + "reward_std": 0.05909878760576248, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4694308340549469, + "rewards/pad": 0.125, + "step": 1753 + }, + { + "completion_length": 69.9375, + "epoch": 0.5589547482472913, + "grad_norm": 19.323455810546875, + "kl": 0.201171875, + "learning_rate": 4.4104525175270876e-07, + "loss": 0.008, + "reward": 1.7000752687454224, + "reward_std": 0.06558647751808167, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.7000752091407776, + "rewards/pad": 0.0, + "step": 1754 + }, + { + "completion_length": 120.6875, + "epoch": 0.5592734225621415, + "grad_norm": 17.894180297851562, + "kl": 0.11962890625, + "learning_rate": 4.407265774378585e-07, + "loss": 0.0048, + "reward": 1.6068882942199707, + "reward_std": 0.05032335966825485, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6068882942199707, + "rewards/pad": 0.0, + "step": 1755 + }, + { + "completion_length": 96.234375, + "epoch": 0.5595920968769917, + "grad_norm": 105.96263885498047, + "kl": 0.11181640625, + "learning_rate": 4.404079031230082e-07, + "loss": 0.0045, + "reward": 1.6398990154266357, + "reward_std": 0.05206108093261719, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5148990750312805, + "rewards/pad": 0.125, + "step": 1756 + }, + { + "completion_length": 96.828125, + "epoch": 0.559910771191842, + "grad_norm": 17.890987396240234, + "kl": 0.1640625, + "learning_rate": 4.40089228808158e-07, + "loss": 0.0066, + "reward": 1.7085633277893066, + "reward_std": 0.11856211721897125, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5991884469985962, + "rewards/pad": 0.109375, + "step": 1757 + }, + { + "completion_length": 119.90625, + "epoch": 0.5602294455066922, + "grad_norm": 37.540985107421875, + "kl": 0.0986328125, + "learning_rate": 4.397705544933078e-07, + "loss": 0.0039, + "reward": 1.4744625091552734, + "reward_std": 0.11637206375598907, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 0.984375, + "rewards/iou_glue_reward": 0.365087628364563, + "step": 1758 + }, + { + "completion_length": 96.984375, + "epoch": 0.5605481198215424, + "grad_norm": 44.76865768432617, + "kl": 0.1982421875, + "learning_rate": 4.394518801784576e-07, + "loss": 0.0079, + "reward": 1.5458180904388428, + "reward_std": 0.139680877327919, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5145681500434875, + "rewards/pad": 0.03125, + "step": 1759 + }, + { + "completion_length": 19.484375, + "epoch": 0.5608667941363926, + "grad_norm": 46.35175704956055, + "kl": 0.2060546875, + "learning_rate": 4.3913320586360734e-07, + "loss": 0.0082, + "reward": 1.5277413129806519, + "reward_std": 0.054538920521736145, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5277412533760071, + "rewards/pad": 0.0, + "step": 1760 + }, + { + "completion_length": 69.8125, + "epoch": 0.5611854684512428, + "grad_norm": 51.44297409057617, + "kl": 0.251953125, + "learning_rate": 4.3881453154875715e-07, + "loss": 0.0101, + "reward": 1.6726279258728027, + "reward_std": 0.08104786276817322, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.547627866268158, + "rewards/pad": 0.125, + "step": 1761 + }, + { + "completion_length": 71.1875, + "epoch": 0.5615041427660931, + "grad_norm": 26.13613510131836, + "kl": 0.15234375, + "learning_rate": 4.384958572339069e-07, + "loss": 0.0061, + "reward": 1.5365188121795654, + "reward_std": 0.05022150278091431, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.2865188717842102, + "rewards/pad": 0.25, + "step": 1762 + }, + { + "completion_length": 95.15625, + "epoch": 0.5618228170809433, + "grad_norm": 15.526792526245117, + "kl": 0.2197265625, + "learning_rate": 4.381771829190567e-07, + "loss": 0.0088, + "reward": 1.558420181274414, + "reward_std": 0.06372374296188354, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4334202706813812, + "rewards/pad": 0.125, + "step": 1763 + }, + { + "completion_length": 46.1875, + "epoch": 0.5621414913957935, + "grad_norm": 38.44749450683594, + "kl": 0.703125, + "learning_rate": 4.3785850860420646e-07, + "loss": 0.0281, + "reward": 1.4561386108398438, + "reward_std": 0.04978647083044052, + "rewards/answer_reward": 0.0, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.456138551235199, + "step": 1764 + }, + { + "completion_length": 97.734375, + "epoch": 0.5624601657106437, + "grad_norm": 26.058134078979492, + "kl": 0.15234375, + "learning_rate": 4.3753983428935627e-07, + "loss": 0.0061, + "reward": 1.730491042137146, + "reward_std": 0.06588118523359299, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.6054909825325012, + "step": 1765 + }, + { + "completion_length": 123.25, + "epoch": 0.5627788400254939, + "grad_norm": 138.15040588378906, + "kl": 0.1005859375, + "learning_rate": 4.37221159974506e-07, + "loss": 0.004, + "reward": 1.4126694202423096, + "reward_std": 0.06151300668716431, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.41266947984695435, + "rewards/pad": 0.0, + "step": 1766 + }, + { + "completion_length": 96.375, + "epoch": 0.5630975143403442, + "grad_norm": 21.328184127807617, + "kl": 0.15625, + "learning_rate": 4.3690248565965583e-07, + "loss": 0.0063, + "reward": 1.5501000881195068, + "reward_std": 0.10059985518455505, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.48760002851486206, + "rewards/pad": 0.0625, + "step": 1767 + }, + { + "completion_length": 98.484375, + "epoch": 0.5634161886551944, + "grad_norm": 46.83586120605469, + "kl": 0.1416015625, + "learning_rate": 4.365838113448056e-07, + "loss": 0.0057, + "reward": 1.8298544883728027, + "reward_std": 0.0899556502699852, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5798544883728027, + "rewards/pad": 0.25, + "step": 1768 + }, + { + "completion_length": 101.953125, + "epoch": 0.5637348629700446, + "grad_norm": 17.387331008911133, + "kl": 0.1298828125, + "learning_rate": 4.362651370299554e-07, + "loss": 0.0052, + "reward": 1.625913381576538, + "reward_std": 0.0826636552810669, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5009133219718933, + "step": 1769 + }, + { + "completion_length": 120.59375, + "epoch": 0.5640535372848948, + "grad_norm": 61.88886260986328, + "kl": 0.291015625, + "learning_rate": 4.3594646271510514e-07, + "loss": 0.0116, + "reward": 1.4453045129776, + "reward_std": 0.11664284765720367, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4453044831752777, + "step": 1770 + }, + { + "completion_length": 97.515625, + "epoch": 0.564372211599745, + "grad_norm": 26.429187774658203, + "kl": 0.10302734375, + "learning_rate": 4.3562778840025495e-07, + "loss": 0.0041, + "reward": 1.506064772605896, + "reward_std": 0.09924240410327911, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4123147130012512, + "rewards/pad": 0.09375, + "step": 1771 + }, + { + "completion_length": 150.515625, + "epoch": 0.5646908859145953, + "grad_norm": 21.579301834106445, + "kl": 0.0927734375, + "learning_rate": 4.353091140854047e-07, + "loss": 0.0037, + "reward": 1.5770467519760132, + "reward_std": 0.09655211120843887, + "rewards/pad": 0.046875, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5301716923713684, + "step": 1772 + }, + { + "completion_length": 150.203125, + "epoch": 0.5650095602294455, + "grad_norm": 15.933002471923828, + "kl": 0.08349609375, + "learning_rate": 4.349904397705545e-07, + "loss": 0.0033, + "reward": 1.4591649770736694, + "reward_std": 0.03933807089924812, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.45916497707366943, + "rewards/pad": 0.0, + "step": 1773 + }, + { + "completion_length": 96.71875, + "epoch": 0.5653282345442957, + "grad_norm": 24.305068969726562, + "kl": 0.1220703125, + "learning_rate": 4.3467176545570427e-07, + "loss": 0.0049, + "reward": 1.490128755569458, + "reward_std": 0.04995978623628616, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.490128755569458, + "rewards/pad": 0.0, + "step": 1774 + }, + { + "completion_length": 96.5625, + "epoch": 0.5656469088591459, + "grad_norm": 47.11887741088867, + "kl": 0.10791015625, + "learning_rate": 4.34353091140854e-07, + "loss": 0.0043, + "reward": 1.543684720993042, + "reward_std": 0.18430109322071075, + "rewards/answer_reward": 0.03125, + "rewards/format_reward_gqa": 0.984375, + "rewards/iou_glue_reward": 0.528059720993042, + "step": 1775 + }, + { + "completion_length": 67.921875, + "epoch": 0.5659655831739961, + "grad_norm": 24.030990600585938, + "kl": 0.19921875, + "learning_rate": 4.340344168260038e-07, + "loss": 0.008, + "reward": 1.4360082149505615, + "reward_std": 0.04120776802301407, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4360082149505615, + "rewards/pad": 0.0, + "step": 1776 + }, + { + "completion_length": 43.765625, + "epoch": 0.5662842574888464, + "grad_norm": 27.24772834777832, + "kl": 0.2412109375, + "learning_rate": 4.337157425111536e-07, + "loss": 0.0096, + "reward": 1.5563490390777588, + "reward_std": 0.04915473610162735, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.556348979473114, + "rewards/pad": 0.0, + "step": 1777 + }, + { + "completion_length": 97.21875, + "epoch": 0.5666029318036966, + "grad_norm": 66.9402084350586, + "kl": 0.1357421875, + "learning_rate": 4.3339706819630333e-07, + "loss": 0.0054, + "reward": 1.52845299243927, + "reward_std": 0.08243995904922485, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4034530520439148, + "rewards/pad": 0.125, + "step": 1778 + }, + { + "completion_length": 70.203125, + "epoch": 0.5669216061185468, + "grad_norm": 42.54038619995117, + "kl": 0.2255859375, + "learning_rate": 4.3307839388145314e-07, + "loss": 0.009, + "reward": 1.5368831157684326, + "reward_std": 0.16717883944511414, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4275081157684326, + "rewards/pad": 0.109375, + "step": 1779 + }, + { + "completion_length": 122.578125, + "epoch": 0.567240280433397, + "grad_norm": 32.484798431396484, + "kl": 0.2001953125, + "learning_rate": 4.327597195666029e-07, + "loss": 0.008, + "reward": 1.7498657703399658, + "reward_std": 0.09915802627801895, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6404908299446106, + "rewards/pad": 0.109375, + "step": 1780 + }, + { + "completion_length": 47.03125, + "epoch": 0.5675589547482472, + "grad_norm": 67.59293365478516, + "kl": 0.1787109375, + "learning_rate": 4.324410452517527e-07, + "loss": 0.0072, + "reward": 1.6660947799682617, + "reward_std": 0.10472682118415833, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5410947203636169, + "step": 1781 + }, + { + "completion_length": 98.328125, + "epoch": 0.5678776290630975, + "grad_norm": 46.848995208740234, + "kl": 0.134765625, + "learning_rate": 4.3212237093690246e-07, + "loss": 0.0054, + "reward": 1.39346182346344, + "reward_std": 0.11564335227012634, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.36221185326576233, + "rewards/pad": 0.03125, + "step": 1782 + }, + { + "completion_length": 174.9375, + "epoch": 0.5681963033779477, + "grad_norm": 6.686620712280273, + "kl": 0.0830078125, + "learning_rate": 4.3180369662205226e-07, + "loss": 0.0033, + "reward": 1.7112741470336914, + "reward_std": 0.1290093958377838, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 0.984375, + "rewards/iou_glue_reward": 0.47689908742904663, + "step": 1783 + }, + { + "completion_length": 68.859375, + "epoch": 0.5685149776927979, + "grad_norm": 33.932472229003906, + "kl": 0.2041015625, + "learning_rate": 4.31485022307202e-07, + "loss": 0.0082, + "reward": 1.681599736213684, + "reward_std": 0.05573929101228714, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5565997362136841, + "step": 1784 + }, + { + "completion_length": 71.453125, + "epoch": 0.5688336520076482, + "grad_norm": 26.65557861328125, + "kl": 0.1572265625, + "learning_rate": 4.3116634799235177e-07, + "loss": 0.0063, + "reward": 1.5919816493988037, + "reward_std": 0.07187262177467346, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5763567686080933, + "rewards/pad": 0.015625, + "step": 1785 + }, + { + "completion_length": 122.90625, + "epoch": 0.5691523263224985, + "grad_norm": 18.70304298400879, + "kl": 0.10986328125, + "learning_rate": 4.308476736775016e-07, + "loss": 0.0044, + "reward": 1.5359832048416138, + "reward_std": 0.07597589492797852, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.535983145236969, + "step": 1786 + }, + { + "completion_length": 121.109375, + "epoch": 0.5694710006373487, + "grad_norm": 27.156719207763672, + "kl": 0.103515625, + "learning_rate": 4.3052899936265133e-07, + "loss": 0.0041, + "reward": 1.549729585647583, + "reward_std": 0.03134915232658386, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4247296452522278, + "step": 1787 + }, + { + "completion_length": 69.9375, + "epoch": 0.5697896749521989, + "grad_norm": 24.777515411376953, + "kl": 0.205078125, + "learning_rate": 4.3021032504780114e-07, + "loss": 0.0082, + "reward": 1.6144177913665771, + "reward_std": 0.08694911748170853, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6144178509712219, + "rewards/pad": 0.0, + "step": 1788 + }, + { + "completion_length": 174.078125, + "epoch": 0.5701083492670491, + "grad_norm": 16.213119506835938, + "kl": 0.11572265625, + "learning_rate": 4.298916507329509e-07, + "loss": 0.0046, + "reward": 1.38360595703125, + "reward_std": 0.09917178004980087, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.3992310166358948, + "step": 1789 + }, + { + "completion_length": 70.078125, + "epoch": 0.5704270235818993, + "grad_norm": 120.56027221679688, + "kl": 0.2353515625, + "learning_rate": 4.295729764181007e-07, + "loss": 0.0094, + "reward": 1.6199557781219482, + "reward_std": 0.11785067617893219, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.49495577812194824, + "step": 1790 + }, + { + "completion_length": 43.453125, + "epoch": 0.5707456978967496, + "grad_norm": 36.520206451416016, + "kl": 0.24609375, + "learning_rate": 4.2925430210325045e-07, + "loss": 0.0099, + "reward": 1.5655598640441895, + "reward_std": 0.06695032119750977, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5655598640441895, + "rewards/pad": 0.0, + "step": 1791 + }, + { + "completion_length": 94.5625, + "epoch": 0.5710643722115998, + "grad_norm": 48.086097717285156, + "kl": 0.1552734375, + "learning_rate": 4.2893562778840026e-07, + "loss": 0.0062, + "reward": 1.5157904624938965, + "reward_std": 0.08498187363147736, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5157905220985413, + "rewards/pad": 0.0, + "step": 1792 + }, + { + "completion_length": 95.015625, + "epoch": 0.57138304652645, + "grad_norm": 16.781322479248047, + "kl": 0.1865234375, + "learning_rate": 4.2861695347355e-07, + "loss": 0.0075, + "reward": 1.5322211980819702, + "reward_std": 0.07854540646076202, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.532221257686615, + "rewards/pad": 0.0, + "step": 1793 + }, + { + "completion_length": 71.96875, + "epoch": 0.5717017208413002, + "grad_norm": 58.54457473754883, + "kl": 0.1435546875, + "learning_rate": 4.282982791586998e-07, + "loss": 0.0057, + "reward": 1.6249440908432007, + "reward_std": 0.1079111397266388, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5780690908432007, + "rewards/pad": 0.046875, + "step": 1794 + }, + { + "completion_length": 71.265625, + "epoch": 0.5720203951561504, + "grad_norm": 17.139041900634766, + "kl": 0.2314453125, + "learning_rate": 4.279796048438495e-07, + "loss": 0.0093, + "reward": 1.4836182594299316, + "reward_std": 0.11333303153514862, + "rewards/answer_reward": 0.0625, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.42111825942993164, + "step": 1795 + }, + { + "completion_length": 42.53125, + "epoch": 0.5723390694710007, + "grad_norm": 37.97492599487305, + "kl": 0.2109375, + "learning_rate": 4.2766093052899933e-07, + "loss": 0.0084, + "reward": 1.445401906967163, + "reward_std": 0.12103722989559174, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4454018473625183, + "rewards/pad": 0.0, + "step": 1796 + }, + { + "completion_length": 124.15625, + "epoch": 0.5726577437858509, + "grad_norm": 35.62852478027344, + "kl": 0.2890625, + "learning_rate": 4.273422562141491e-07, + "loss": 0.0115, + "reward": 1.4103597402572632, + "reward_std": 0.09135589748620987, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.30098479986190796, + "rewards/pad": 0.125, + "step": 1797 + }, + { + "completion_length": 123.625, + "epoch": 0.5729764181007011, + "grad_norm": 31.406539916992188, + "kl": 0.146484375, + "learning_rate": 4.270235818992989e-07, + "loss": 0.0058, + "reward": 1.5586957931518555, + "reward_std": 0.0993969589471817, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5586957931518555, + "rewards/pad": 0.0, + "step": 1798 + }, + { + "completion_length": 97.75, + "epoch": 0.5732950924155513, + "grad_norm": 28.93682098388672, + "kl": 0.15234375, + "learning_rate": 4.2670490758444865e-07, + "loss": 0.0061, + "reward": 1.4915449619293213, + "reward_std": 0.08422195911407471, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4759199023246765, + "rewards/pad": 0.015625, + "step": 1799 + }, + { + "completion_length": 68.421875, + "epoch": 0.5736137667304015, + "grad_norm": 265.1728210449219, + "kl": 0.2216796875, + "learning_rate": 4.2638623326959845e-07, + "loss": 0.0089, + "reward": 1.5189340114593506, + "reward_std": 0.10375507175922394, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5189339518547058, + "rewards/pad": 0.0, + "step": 1800 + }, + { + "completion_length": 172.65625, + "epoch": 0.5739324410452518, + "grad_norm": 8.524855613708496, + "kl": 0.08984375, + "learning_rate": 4.260675589547482e-07, + "loss": 0.0036, + "reward": 1.3667278289794922, + "reward_std": 0.03605463728308678, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3667277693748474, + "step": 1801 + }, + { + "completion_length": 97.78125, + "epoch": 0.574251115360102, + "grad_norm": 28.816625595092773, + "kl": 0.1572265625, + "learning_rate": 4.25748884639898e-07, + "loss": 0.0063, + "reward": 1.5991346836090088, + "reward_std": 0.05232204869389534, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.34913477301597595, + "step": 1802 + }, + { + "completion_length": 18.375, + "epoch": 0.5745697896749522, + "grad_norm": 62.05613327026367, + "kl": 0.189453125, + "learning_rate": 4.2543021032504777e-07, + "loss": 0.0076, + "reward": 1.6871989965438843, + "reward_std": 0.08016446232795715, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6871989965438843, + "rewards/pad": 0.0, + "step": 1803 + }, + { + "completion_length": 71.40625, + "epoch": 0.5748884639898024, + "grad_norm": 20.37428855895996, + "kl": 0.283203125, + "learning_rate": 4.251115360101976e-07, + "loss": 0.0114, + "reward": 1.819273829460144, + "reward_std": 0.08338651806116104, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.569273829460144, + "rewards/pad": 0.25, + "step": 1804 + }, + { + "completion_length": 95.609375, + "epoch": 0.5752071383046526, + "grad_norm": 34.7626953125, + "kl": 0.14453125, + "learning_rate": 4.2479286169534733e-07, + "loss": 0.0058, + "reward": 1.725312352180481, + "reward_std": 0.04575943201780319, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.600312352180481, + "rewards/pad": 0.125, + "step": 1805 + }, + { + "completion_length": 99.5625, + "epoch": 0.5755258126195029, + "grad_norm": 44.738346099853516, + "kl": 0.18359375, + "learning_rate": 4.2447418738049714e-07, + "loss": 0.0074, + "reward": 1.5814894437789917, + "reward_std": 0.10402372479438782, + "rewards/answer_reward": 0.15625, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4252393841743469, + "step": 1806 + }, + { + "completion_length": 122.421875, + "epoch": 0.5758444869343531, + "grad_norm": 38.684932708740234, + "kl": 0.08837890625, + "learning_rate": 4.241555130656469e-07, + "loss": 0.0035, + "reward": 1.5361063480377197, + "reward_std": 0.0688786506652832, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5361063480377197, + "rewards/pad": 0.0, + "step": 1807 + }, + { + "completion_length": 97.640625, + "epoch": 0.5761631612492033, + "grad_norm": 117.1318130493164, + "kl": 0.28125, + "learning_rate": 4.238368387507967e-07, + "loss": 0.0112, + "reward": 1.5856540203094482, + "reward_std": 0.10018368065357208, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.570029079914093, + "rewards/pad": 0.015625, + "step": 1808 + }, + { + "completion_length": 172.546875, + "epoch": 0.5764818355640535, + "grad_norm": 19.1522216796875, + "kl": 0.09814453125, + "learning_rate": 4.2351816443594645e-07, + "loss": 0.0039, + "reward": 1.5147862434387207, + "reward_std": 0.03550882637500763, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5147862434387207, + "rewards/pad": 0.0, + "step": 1809 + }, + { + "completion_length": 95.53125, + "epoch": 0.5768005098789037, + "grad_norm": 53.894081115722656, + "kl": 0.1435546875, + "learning_rate": 4.2319949012109626e-07, + "loss": 0.0057, + "reward": 1.4878883361816406, + "reward_std": 0.05823700875043869, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4878883361816406, + "rewards/pad": 0.0, + "step": 1810 + }, + { + "completion_length": 122.390625, + "epoch": 0.577119184193754, + "grad_norm": 48.407161712646484, + "kl": 0.15625, + "learning_rate": 4.22880815806246e-07, + "loss": 0.0062, + "reward": 1.5016330480575562, + "reward_std": 0.08562350273132324, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5016330480575562, + "rewards/pad": 0.0, + "step": 1811 + }, + { + "completion_length": 70.59375, + "epoch": 0.5774378585086042, + "grad_norm": 52.78874206542969, + "kl": 0.1962890625, + "learning_rate": 4.225621414913958e-07, + "loss": 0.0079, + "reward": 1.5229122638702393, + "reward_std": 0.09713631123304367, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5229122042655945, + "rewards/pad": 0.0, + "step": 1812 + }, + { + "completion_length": 44.734375, + "epoch": 0.5777565328234544, + "grad_norm": 31.73822021484375, + "kl": 0.2275390625, + "learning_rate": 4.2224346717654557e-07, + "loss": 0.0091, + "reward": 1.7680362462997437, + "reward_std": 0.08835802972316742, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6430363655090332, + "rewards/pad": 0.125, + "step": 1813 + }, + { + "completion_length": 96.34375, + "epoch": 0.5780752071383046, + "grad_norm": 42.44234848022461, + "kl": 0.1318359375, + "learning_rate": 4.219247928616954e-07, + "loss": 0.0053, + "reward": 1.6047329902648926, + "reward_std": 0.1471775770187378, + "rewards/answer_reward": 0.09375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5109829902648926, + "step": 1814 + }, + { + "completion_length": 121.953125, + "epoch": 0.5783938814531548, + "grad_norm": 57.97667694091797, + "kl": 0.11279296875, + "learning_rate": 4.216061185468451e-07, + "loss": 0.0045, + "reward": 1.438378930091858, + "reward_std": 0.1234152764081955, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.4540039300918579, + "rewards/pad": 0.0, + "step": 1815 + }, + { + "completion_length": 97.984375, + "epoch": 0.5787125557680051, + "grad_norm": 25.46637535095215, + "kl": 0.142578125, + "learning_rate": 4.2128744423199483e-07, + "loss": 0.0057, + "reward": 1.6345410346984863, + "reward_std": 0.05224389582872391, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5095410346984863, + "step": 1816 + }, + { + "completion_length": 122.8125, + "epoch": 0.5790312300828553, + "grad_norm": 27.115081787109375, + "kl": 0.162109375, + "learning_rate": 4.2096876991714464e-07, + "loss": 0.0065, + "reward": 1.7935987710952759, + "reward_std": 0.05079444497823715, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5435987114906311, + "rewards/pad": 0.25, + "step": 1817 + }, + { + "completion_length": 96.640625, + "epoch": 0.5793499043977055, + "grad_norm": 28.72382926940918, + "kl": 0.1494140625, + "learning_rate": 4.206500956022944e-07, + "loss": 0.006, + "reward": 1.5233402252197266, + "reward_std": 0.10524062067270279, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3983403444290161, + "rewards/pad": 0.125, + "step": 1818 + }, + { + "completion_length": 97.015625, + "epoch": 0.5796685787125557, + "grad_norm": 13.208295822143555, + "kl": 0.1806640625, + "learning_rate": 4.203314212874442e-07, + "loss": 0.0072, + "reward": 1.5535154342651367, + "reward_std": 0.09736257791519165, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.42851537466049194, + "rewards/pad": 0.125, + "step": 1819 + }, + { + "completion_length": 69.984375, + "epoch": 0.579987253027406, + "grad_norm": 65.50917053222656, + "kl": 0.1953125, + "learning_rate": 4.2001274697259396e-07, + "loss": 0.0078, + "reward": 1.6009665727615356, + "reward_std": 0.048545848578214645, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6009665727615356, + "rewards/pad": 0.0, + "step": 1820 + }, + { + "completion_length": 97.625, + "epoch": 0.5803059273422562, + "grad_norm": 75.36929321289062, + "kl": 0.12158203125, + "learning_rate": 4.1969407265774376e-07, + "loss": 0.0049, + "reward": 1.6634126901626587, + "reward_std": 0.06822207570075989, + "rewards/answer_reward": 0.234375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4290376901626587, + "step": 1821 + }, + { + "completion_length": 198.265625, + "epoch": 0.5806246016571064, + "grad_norm": 6.578020095825195, + "kl": 0.06640625, + "learning_rate": 4.193753983428935e-07, + "loss": 0.0026, + "reward": 1.4357388019561768, + "reward_std": 0.09460680186748505, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.45136383175849915, + "step": 1822 + }, + { + "completion_length": 95.703125, + "epoch": 0.5809432759719566, + "grad_norm": 50.63115310668945, + "kl": 0.142578125, + "learning_rate": 4.190567240280433e-07, + "loss": 0.0057, + "reward": 1.6899892091751099, + "reward_std": 0.05263391509652138, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5649892091751099, + "step": 1823 + }, + { + "completion_length": 94.921875, + "epoch": 0.5812619502868069, + "grad_norm": 44.11355209350586, + "kl": 0.26953125, + "learning_rate": 4.187380497131931e-07, + "loss": 0.0108, + "reward": 1.6089653968811035, + "reward_std": 0.09958036243915558, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.48396536707878113, + "step": 1824 + }, + { + "completion_length": 99.0, + "epoch": 0.5815806246016572, + "grad_norm": 17.71644401550293, + "kl": 0.11181640625, + "learning_rate": 4.184193753983429e-07, + "loss": 0.0045, + "reward": 1.6234378814697266, + "reward_std": 0.03895140066742897, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3734378516674042, + "rewards/pad": 0.25, + "step": 1825 + }, + { + "completion_length": 70.234375, + "epoch": 0.5818992989165074, + "grad_norm": 22.119213104248047, + "kl": 0.154296875, + "learning_rate": 4.1810070108349264e-07, + "loss": 0.0062, + "reward": 1.7264617681503296, + "reward_std": 0.08063302934169769, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6014617681503296, + "rewards/pad": 0.125, + "step": 1826 + }, + { + "completion_length": 96.515625, + "epoch": 0.5822179732313576, + "grad_norm": 19.158109664916992, + "kl": 0.2216796875, + "learning_rate": 4.1778202676864245e-07, + "loss": 0.0089, + "reward": 1.5653225183486938, + "reward_std": 0.11509357392787933, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5496974587440491, + "rewards/pad": 0.015625, + "step": 1827 + }, + { + "completion_length": 71.265625, + "epoch": 0.5825366475462078, + "grad_norm": 45.197513580322266, + "kl": 0.1435546875, + "learning_rate": 4.174633524537922e-07, + "loss": 0.0057, + "reward": 1.6790035963058472, + "reward_std": 0.13340117037296295, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5696284770965576, + "rewards/pad": 0.109375, + "step": 1828 + }, + { + "completion_length": 119.9375, + "epoch": 0.582855321861058, + "grad_norm": 15.074068069458008, + "kl": 0.2197265625, + "learning_rate": 4.17144678138942e-07, + "loss": 0.0088, + "reward": 1.643786907196045, + "reward_std": 0.06817853450775146, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6437869071960449, + "rewards/pad": 0.0, + "step": 1829 + }, + { + "completion_length": 67.296875, + "epoch": 0.5831739961759083, + "grad_norm": 94.0682601928711, + "kl": 0.2138671875, + "learning_rate": 4.1682600382409176e-07, + "loss": 0.0086, + "reward": 1.5750195980072021, + "reward_std": 0.1432129293680191, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5750196576118469, + "step": 1830 + }, + { + "completion_length": 68.46875, + "epoch": 0.5834926704907585, + "grad_norm": 38.54087829589844, + "kl": 0.1435546875, + "learning_rate": 4.1650732950924157e-07, + "loss": 0.0058, + "reward": 1.5484356880187988, + "reward_std": 0.0928826630115509, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5484358072280884, + "rewards/pad": 0.0, + "step": 1831 + }, + { + "completion_length": 97.8125, + "epoch": 0.5838113448056087, + "grad_norm": 19.759061813354492, + "kl": 0.10302734375, + "learning_rate": 4.161886551943913e-07, + "loss": 0.0041, + "reward": 1.604832410812378, + "reward_std": 0.09509650617837906, + "rewards/pad": 0.015625, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5892074108123779, + "step": 1832 + }, + { + "completion_length": 149.71875, + "epoch": 0.5841300191204589, + "grad_norm": 22.152210235595703, + "kl": 0.0859375, + "learning_rate": 4.1586998087954113e-07, + "loss": 0.0034, + "reward": 1.5515859127044678, + "reward_std": 0.11550014466047287, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.44221100211143494, + "step": 1833 + }, + { + "completion_length": 119.109375, + "epoch": 0.5844486934353091, + "grad_norm": 56.19287109375, + "kl": 0.1962890625, + "learning_rate": 4.1555130656469083e-07, + "loss": 0.0078, + "reward": 1.4814128875732422, + "reward_std": 0.0679532140493393, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.48141294717788696, + "rewards/pad": 0.0, + "step": 1834 + }, + { + "completion_length": 169.265625, + "epoch": 0.5847673677501594, + "grad_norm": 386.7119445800781, + "kl": 0.1533203125, + "learning_rate": 4.1523263224984064e-07, + "loss": 0.0061, + "reward": 1.359937071800232, + "reward_std": 0.09846100956201553, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.35993701219558716, + "step": 1835 + }, + { + "completion_length": 143.734375, + "epoch": 0.5850860420650096, + "grad_norm": 30.239673614501953, + "kl": 0.1572265625, + "learning_rate": 4.149139579349904e-07, + "loss": 0.0063, + "reward": 1.6132328510284424, + "reward_std": 0.06773644685745239, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4882327914237976, + "step": 1836 + }, + { + "completion_length": 93.5625, + "epoch": 0.5854047163798598, + "grad_norm": 36.47754669189453, + "kl": 0.173828125, + "learning_rate": 4.145952836201402e-07, + "loss": 0.0069, + "reward": 1.5574928522109985, + "reward_std": 0.06488293409347534, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5574928522109985, + "rewards/pad": 0.0, + "step": 1837 + }, + { + "completion_length": 121.03125, + "epoch": 0.58572339069471, + "grad_norm": 13.884906768798828, + "kl": 0.076171875, + "learning_rate": 4.1427660930528995e-07, + "loss": 0.003, + "reward": 1.6176261901855469, + "reward_std": 0.04010601341724396, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.49262624979019165, + "step": 1838 + }, + { + "completion_length": 67.0, + "epoch": 0.5860420650095602, + "grad_norm": 24.440515518188477, + "kl": 0.2158203125, + "learning_rate": 4.1395793499043976e-07, + "loss": 0.0086, + "reward": 1.5812326669692993, + "reward_std": 0.08772775530815125, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5812326669692993, + "step": 1839 + }, + { + "completion_length": 95.125, + "epoch": 0.5863607393244105, + "grad_norm": 39.3325309753418, + "kl": 0.11669921875, + "learning_rate": 4.136392606755895e-07, + "loss": 0.0047, + "reward": 1.7650656700134277, + "reward_std": 0.06125748157501221, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6400656700134277, + "rewards/pad": 0.125, + "step": 1840 + }, + { + "completion_length": 118.09375, + "epoch": 0.5866794136392607, + "grad_norm": 9.982199668884277, + "kl": 0.1513671875, + "learning_rate": 4.133205863607393e-07, + "loss": 0.006, + "reward": 1.61601984500885, + "reward_std": 0.04292646422982216, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6160199046134949, + "rewards/pad": 0.0, + "step": 1841 + }, + { + "completion_length": 98.453125, + "epoch": 0.5869980879541109, + "grad_norm": 26.047142028808594, + "kl": 0.2177734375, + "learning_rate": 4.130019120458891e-07, + "loss": 0.0087, + "reward": 1.7720967531204224, + "reward_std": 0.0968465581536293, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5220966339111328, + "step": 1842 + }, + { + "completion_length": 119.5625, + "epoch": 0.5873167622689611, + "grad_norm": 33.48945617675781, + "kl": 0.09912109375, + "learning_rate": 4.126832377310389e-07, + "loss": 0.004, + "reward": 1.5689330101013184, + "reward_std": 0.04028594493865967, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5689329504966736, + "rewards/pad": 0.0, + "step": 1843 + }, + { + "completion_length": 97.828125, + "epoch": 0.5876354365838113, + "grad_norm": 79.5909194946289, + "kl": 0.134765625, + "learning_rate": 4.1236456341618864e-07, + "loss": 0.0054, + "reward": 1.5185623168945312, + "reward_std": 0.1128365620970726, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.37793731689453125, + "rewards/pad": 0.140625, + "step": 1844 + }, + { + "completion_length": 70.90625, + "epoch": 0.5879541108986616, + "grad_norm": 18.25775146484375, + "kl": 0.2197265625, + "learning_rate": 4.1204588910133844e-07, + "loss": 0.0088, + "reward": 1.4915564060211182, + "reward_std": 0.140813410282135, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.5071814656257629, + "step": 1845 + }, + { + "completion_length": 146.21875, + "epoch": 0.5882727852135118, + "grad_norm": 6.8526716232299805, + "kl": 0.1005859375, + "learning_rate": 4.117272147864882e-07, + "loss": 0.004, + "reward": 1.633963942527771, + "reward_std": 0.04471247270703316, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.508963942527771, + "step": 1846 + }, + { + "completion_length": 96.828125, + "epoch": 0.588591459528362, + "grad_norm": 11.51022720336914, + "kl": 0.1572265625, + "learning_rate": 4.1140854047163795e-07, + "loss": 0.0063, + "reward": 1.5092898607254028, + "reward_std": 0.05117335915565491, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5092898607254028, + "step": 1847 + }, + { + "completion_length": 70.953125, + "epoch": 0.5889101338432122, + "grad_norm": 28.110187530517578, + "kl": 0.1591796875, + "learning_rate": 4.1108986615678776e-07, + "loss": 0.0064, + "reward": 1.651709794998169, + "reward_std": 0.06477080285549164, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.651709794998169, + "step": 1848 + }, + { + "completion_length": 145.328125, + "epoch": 0.5892288081580624, + "grad_norm": 20.159751892089844, + "kl": 0.119140625, + "learning_rate": 4.107711918419375e-07, + "loss": 0.0048, + "reward": 1.5828752517700195, + "reward_std": 0.07095219939947128, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4578753411769867, + "step": 1849 + }, + { + "completion_length": 121.828125, + "epoch": 0.5895474824729127, + "grad_norm": 421.111572265625, + "kl": 0.125, + "learning_rate": 4.104525175270873e-07, + "loss": 0.005, + "reward": 1.7188516855239868, + "reward_std": 0.06404486298561096, + "rewards/pad": 0.25, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4688516855239868, + "step": 1850 + }, + { + "completion_length": 100.515625, + "epoch": 0.5898661567877629, + "grad_norm": 14.096718788146973, + "kl": 0.1611328125, + "learning_rate": 4.1013384321223707e-07, + "loss": 0.0064, + "reward": 1.4676356315612793, + "reward_std": 0.07982802391052246, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3426356315612793, + "rewards/pad": 0.125, + "step": 1851 + }, + { + "completion_length": 97.59375, + "epoch": 0.5901848311026131, + "grad_norm": 31.308773040771484, + "kl": 0.1376953125, + "learning_rate": 4.098151688973869e-07, + "loss": 0.0055, + "reward": 1.7079538106918335, + "reward_std": 0.1335076093673706, + "rewards/pad": 0.140625, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5673288106918335, + "step": 1852 + }, + { + "completion_length": 118.609375, + "epoch": 0.5905035054174633, + "grad_norm": 28.14283561706543, + "kl": 0.1337890625, + "learning_rate": 4.0949649458253663e-07, + "loss": 0.0053, + "reward": 1.4441075325012207, + "reward_std": 0.06789509207010269, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4441075325012207, + "rewards/pad": 0.0, + "step": 1853 + }, + { + "completion_length": 95.78125, + "epoch": 0.5908221797323135, + "grad_norm": 20.888446807861328, + "kl": 0.1435546875, + "learning_rate": 4.091778202676864e-07, + "loss": 0.0057, + "reward": 1.7055909633636475, + "reward_std": 0.11922048777341843, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.5962159037590027, + "rewards/pad": 0.125, + "step": 1854 + }, + { + "completion_length": 45.734375, + "epoch": 0.5911408540471638, + "grad_norm": 49.40755844116211, + "kl": 0.197265625, + "learning_rate": 4.0885914595283614e-07, + "loss": 0.0079, + "reward": 1.4927027225494385, + "reward_std": 0.22029918432235718, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3520776927471161, + "rewards/pad": 0.140625, + "step": 1855 + }, + { + "completion_length": 93.4375, + "epoch": 0.591459528362014, + "grad_norm": 51.1677131652832, + "kl": 0.1650390625, + "learning_rate": 4.0854047163798595e-07, + "loss": 0.0066, + "reward": 1.5748016834259033, + "reward_std": 0.07330948114395142, + "rewards/answer_reward": 0.0, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5748017430305481, + "step": 1856 + }, + { + "completion_length": 70.703125, + "epoch": 0.5917782026768642, + "grad_norm": 38.13994598388672, + "kl": 0.1826171875, + "learning_rate": 4.082217973231357e-07, + "loss": 0.0073, + "reward": 1.5587074756622314, + "reward_std": 0.07806047797203064, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5587073564529419, + "step": 1857 + }, + { + "completion_length": 71.9375, + "epoch": 0.5920968769917144, + "grad_norm": 18.738344192504883, + "kl": 0.1533203125, + "learning_rate": 4.079031230082855e-07, + "loss": 0.0062, + "reward": 1.5912575721740723, + "reward_std": 0.1442650705575943, + "rewards/answer_reward": 0.0625, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5287575721740723, + "step": 1858 + }, + { + "completion_length": 70.90625, + "epoch": 0.5924155513065646, + "grad_norm": 56.776798248291016, + "kl": 0.201171875, + "learning_rate": 4.0758444869343526e-07, + "loss": 0.0081, + "reward": 1.5838427543640137, + "reward_std": 0.06699131429195404, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5838428139686584, + "step": 1859 + }, + { + "completion_length": 96.625, + "epoch": 0.5927342256214149, + "grad_norm": 32.76490783691406, + "kl": 0.1318359375, + "learning_rate": 4.0726577437858507e-07, + "loss": 0.0053, + "reward": 1.5736356973648071, + "reward_std": 0.09358209371566772, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4486357271671295, + "rewards/pad": 0.125, + "step": 1860 + }, + { + "completion_length": 44.96875, + "epoch": 0.5930528999362651, + "grad_norm": 28.009550094604492, + "kl": 0.359375, + "learning_rate": 4.069471000637348e-07, + "loss": 0.0144, + "reward": 1.483938455581665, + "reward_std": 0.12422734498977661, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4526885151863098, + "rewards/pad": 0.03125, + "step": 1861 + }, + { + "completion_length": 122.171875, + "epoch": 0.5933715742511153, + "grad_norm": 17.696401596069336, + "kl": 0.1337890625, + "learning_rate": 4.0662842574888463e-07, + "loss": 0.0054, + "reward": 1.4019179344177246, + "reward_std": 0.05144944787025452, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.40191787481307983, + "rewards/pad": 0.0, + "step": 1862 + }, + { + "completion_length": 120.59375, + "epoch": 0.5936902485659655, + "grad_norm": 28.02484703063965, + "kl": 0.1103515625, + "learning_rate": 4.063097514340344e-07, + "loss": 0.0044, + "reward": 1.4743096828460693, + "reward_std": 0.06880053877830505, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4743097126483917, + "step": 1863 + }, + { + "completion_length": 19.40625, + "epoch": 0.5940089228808159, + "grad_norm": 44.00604248046875, + "kl": 0.359375, + "learning_rate": 4.059910771191842e-07, + "loss": 0.0144, + "reward": 1.7003247737884521, + "reward_std": 0.10178287327289581, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5753247737884521, + "rewards/pad": 0.125, + "step": 1864 + }, + { + "completion_length": 98.078125, + "epoch": 0.5943275971956661, + "grad_norm": 22.15548324584961, + "kl": 0.1533203125, + "learning_rate": 4.0567240280433395e-07, + "loss": 0.0061, + "reward": 1.7335445880889893, + "reward_std": 0.10923653095960617, + "rewards/pad": 0.21875, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5147947072982788, + "step": 1865 + }, + { + "completion_length": 121.0, + "epoch": 0.5946462715105163, + "grad_norm": 11.890228271484375, + "kl": 0.126953125, + "learning_rate": 4.0535372848948375e-07, + "loss": 0.0051, + "reward": 1.6787242889404297, + "reward_std": 0.07239396870136261, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.6787242889404297, + "step": 1866 + }, + { + "completion_length": 71.21875, + "epoch": 0.5949649458253665, + "grad_norm": 42.6705322265625, + "kl": 0.1357421875, + "learning_rate": 4.050350541746335e-07, + "loss": 0.0054, + "reward": 1.7034825086593628, + "reward_std": 0.11921627819538116, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.7191076278686523, + "step": 1867 + }, + { + "completion_length": 122.421875, + "epoch": 0.5952836201402167, + "grad_norm": 11.307313919067383, + "kl": 0.1923828125, + "learning_rate": 4.047163798597833e-07, + "loss": 0.0077, + "reward": 1.6598554849624634, + "reward_std": 0.055303771048784256, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.40985557436943054, + "rewards/pad": 0.25, + "step": 1868 + }, + { + "completion_length": 45.5, + "epoch": 0.595602294455067, + "grad_norm": 36.23927307128906, + "kl": 0.1865234375, + "learning_rate": 4.0439770554493307e-07, + "loss": 0.0074, + "reward": 1.6517199277877808, + "reward_std": 0.14537964761257172, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.5423449277877808, + "step": 1869 + }, + { + "completion_length": 122.71875, + "epoch": 0.5959209687699172, + "grad_norm": 68.48987579345703, + "kl": 0.1298828125, + "learning_rate": 4.040790312300829e-07, + "loss": 0.0052, + "reward": 1.3829237222671509, + "reward_std": 0.1671084463596344, + "rewards/pad": 0.0625, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.3360486924648285, + "step": 1870 + }, + { + "completion_length": 93.859375, + "epoch": 0.5962396430847674, + "grad_norm": 73.4350357055664, + "kl": 0.2216796875, + "learning_rate": 4.0376035691523263e-07, + "loss": 0.0089, + "reward": 1.7129826545715332, + "reward_std": 0.11085135489702225, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5879825949668884, + "step": 1871 + }, + { + "completion_length": 121.109375, + "epoch": 0.5965583173996176, + "grad_norm": 10.7921724319458, + "kl": 0.203125, + "learning_rate": 4.0344168260038244e-07, + "loss": 0.0082, + "reward": 1.5458829402923584, + "reward_std": 0.05583937466144562, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4208829998970032, + "rewards/pad": 0.125, + "step": 1872 + }, + { + "completion_length": 146.875, + "epoch": 0.5968769917144678, + "grad_norm": 14.976286888122559, + "kl": 0.0986328125, + "learning_rate": 4.0312300828553214e-07, + "loss": 0.0039, + "reward": 1.5083661079406738, + "reward_std": 0.05651112645864487, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5083661675453186, + "step": 1873 + }, + { + "completion_length": 46.0625, + "epoch": 0.5971956660293181, + "grad_norm": 99.23194122314453, + "kl": 0.177734375, + "learning_rate": 4.0280433397068195e-07, + "loss": 0.0071, + "reward": 1.5729570388793945, + "reward_std": 0.16992275416851044, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.44795697927474976, + "rewards/pad": 0.125, + "step": 1874 + }, + { + "completion_length": 69.484375, + "epoch": 0.5975143403441683, + "grad_norm": 12.070338249206543, + "kl": 0.1484375, + "learning_rate": 4.024856596558317e-07, + "loss": 0.0059, + "reward": 1.510481357574463, + "reward_std": 0.04901345819234848, + "rewards/answer_reward": 0.0, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5104814171791077, + "step": 1875 + }, + { + "completion_length": 123.3125, + "epoch": 0.5978330146590185, + "grad_norm": 76.7409439086914, + "kl": 0.09814453125, + "learning_rate": 4.0216698534098145e-07, + "loss": 0.0039, + "reward": 1.4769536256790161, + "reward_std": 0.08186715096235275, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3675785958766937, + "rewards/pad": 0.109375, + "step": 1876 + }, + { + "completion_length": 46.25, + "epoch": 0.5981516889738687, + "grad_norm": 47.11528396606445, + "kl": 0.19140625, + "learning_rate": 4.0184831102613126e-07, + "loss": 0.0077, + "reward": 1.580622673034668, + "reward_std": 0.13584300875663757, + "rewards/answer_reward": 0.15625, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4243726134300232, + "step": 1877 + }, + { + "completion_length": 70.234375, + "epoch": 0.5984703632887189, + "grad_norm": 39.12364959716797, + "kl": 0.2890625, + "learning_rate": 4.01529636711281e-07, + "loss": 0.0115, + "reward": 1.7271225452423096, + "reward_std": 0.11779257655143738, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6177475452423096, + "rewards/pad": 0.109375, + "step": 1878 + }, + { + "completion_length": 43.578125, + "epoch": 0.5987890376035692, + "grad_norm": 39.20248031616211, + "kl": 0.27734375, + "learning_rate": 4.012109623964308e-07, + "loss": 0.0111, + "reward": 1.6702015399932861, + "reward_std": 0.08535203337669373, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6702016592025757, + "rewards/pad": 0.0, + "step": 1879 + }, + { + "completion_length": 122.34375, + "epoch": 0.5991077119184194, + "grad_norm": 34.952552795410156, + "kl": 0.11328125, + "learning_rate": 4.008922880815806e-07, + "loss": 0.0045, + "reward": 1.4387636184692383, + "reward_std": 0.04700171947479248, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.43876367807388306, + "rewards/pad": 0.0, + "step": 1880 + }, + { + "completion_length": 71.3125, + "epoch": 0.5994263862332696, + "grad_norm": 30.318069458007812, + "kl": 0.1259765625, + "learning_rate": 4.005736137667304e-07, + "loss": 0.005, + "reward": 1.5658936500549316, + "reward_std": 0.05282466858625412, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.44089367985725403, + "step": 1881 + }, + { + "completion_length": 94.4375, + "epoch": 0.5997450605481198, + "grad_norm": 12.464398384094238, + "kl": 0.181640625, + "learning_rate": 4.0025493945188014e-07, + "loss": 0.0073, + "reward": 1.4137916564941406, + "reward_std": 0.07801970094442368, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4137916564941406, + "rewards/pad": 0.0, + "step": 1882 + }, + { + "completion_length": 71.21875, + "epoch": 0.60006373486297, + "grad_norm": 67.10906982421875, + "kl": 0.1650390625, + "learning_rate": 3.9993626513702994e-07, + "loss": 0.0066, + "reward": 1.7139067649841309, + "reward_std": 0.11026249825954437, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4951567053794861, + "rewards/pad": 0.21875, + "step": 1883 + }, + { + "completion_length": 98.40625, + "epoch": 0.6003824091778203, + "grad_norm": 29.317354202270508, + "kl": 0.14453125, + "learning_rate": 3.996175908221797e-07, + "loss": 0.0058, + "reward": 1.4955264329910278, + "reward_std": 0.06369823217391968, + "rewards/answer_reward": 0.0, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.49552643299102783, + "step": 1884 + }, + { + "completion_length": 144.078125, + "epoch": 0.6007010834926705, + "grad_norm": 20.339582443237305, + "kl": 0.1142578125, + "learning_rate": 3.992989165073295e-07, + "loss": 0.0046, + "reward": 1.5659692287445068, + "reward_std": 0.11166795343160629, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.48784422874450684, + "rewards/pad": 0.078125, + "step": 1885 + }, + { + "completion_length": 67.4375, + "epoch": 0.6010197578075207, + "grad_norm": 66.48014831542969, + "kl": 0.1630859375, + "learning_rate": 3.9898024219247926e-07, + "loss": 0.0065, + "reward": 1.6489007472991943, + "reward_std": 0.11646482348442078, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6489008665084839, + "rewards/pad": 0.0, + "step": 1886 + }, + { + "completion_length": 122.890625, + "epoch": 0.6013384321223709, + "grad_norm": 35.8299560546875, + "kl": 0.09619140625, + "learning_rate": 3.9866156787762907e-07, + "loss": 0.0039, + "reward": 1.6577228307724, + "reward_std": 0.06755958497524261, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5327227711677551, + "step": 1887 + }, + { + "completion_length": 43.4375, + "epoch": 0.6016571064372211, + "grad_norm": 18.268341064453125, + "kl": 0.1474609375, + "learning_rate": 3.983428935627788e-07, + "loss": 0.0059, + "reward": 1.5191141366958618, + "reward_std": 0.050555162131786346, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5191141366958618, + "step": 1888 + }, + { + "completion_length": 97.578125, + "epoch": 0.6019757807520714, + "grad_norm": 21.400203704833984, + "kl": 0.1669921875, + "learning_rate": 3.980242192479286e-07, + "loss": 0.0067, + "reward": 1.8046271800994873, + "reward_std": 0.16014282405376434, + "rewards/answer_reward": 0.296875, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5077521800994873, + "step": 1889 + }, + { + "completion_length": 44.578125, + "epoch": 0.6022944550669216, + "grad_norm": 42.205665588378906, + "kl": 0.609375, + "learning_rate": 3.977055449330784e-07, + "loss": 0.0244, + "reward": 1.6509807109832764, + "reward_std": 0.11738397926092148, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6509807705879211, + "rewards/pad": 0.0, + "step": 1890 + }, + { + "completion_length": 122.078125, + "epoch": 0.6026131293817718, + "grad_norm": 23.141324996948242, + "kl": 0.10986328125, + "learning_rate": 3.973868706182282e-07, + "loss": 0.0044, + "reward": 1.5159111022949219, + "reward_std": 0.09590984135866165, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3909110724925995, + "step": 1891 + }, + { + "completion_length": 121.40625, + "epoch": 0.602931803696622, + "grad_norm": 20.430015563964844, + "kl": 0.087890625, + "learning_rate": 3.9706819630337794e-07, + "loss": 0.0035, + "reward": 1.6643826961517334, + "reward_std": 0.09546613693237305, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 0.984375, + "rewards/iou_glue_reward": 0.5550077557563782, + "step": 1892 + }, + { + "completion_length": 44.4375, + "epoch": 0.6032504780114722, + "grad_norm": 54.14018249511719, + "kl": 0.16015625, + "learning_rate": 3.967495219885277e-07, + "loss": 0.0064, + "reward": 1.2324488162994385, + "reward_std": 0.07691571861505508, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.2324487566947937, + "step": 1893 + }, + { + "completion_length": 45.84375, + "epoch": 0.6035691523263225, + "grad_norm": 68.19710540771484, + "kl": 0.23046875, + "learning_rate": 3.9643084767367745e-07, + "loss": 0.0092, + "reward": 1.6674814224243164, + "reward_std": 0.10494668036699295, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.526856541633606, + "rewards/pad": 0.140625, + "step": 1894 + }, + { + "completion_length": 171.859375, + "epoch": 0.6038878266411727, + "grad_norm": 10.1597318649292, + "kl": 0.080078125, + "learning_rate": 3.9611217335882726e-07, + "loss": 0.0032, + "reward": 1.592544436454773, + "reward_std": 0.053140126168727875, + "rewards/answer_reward": 0.125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4675444960594177, + "step": 1895 + }, + { + "completion_length": 92.71875, + "epoch": 0.6042065009560229, + "grad_norm": 21.789133071899414, + "kl": 0.2353515625, + "learning_rate": 3.95793499043977e-07, + "loss": 0.0094, + "reward": 1.6179006099700928, + "reward_std": 0.0610184520483017, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6179004907608032, + "rewards/pad": 0.0, + "step": 1896 + }, + { + "completion_length": 45.96875, + "epoch": 0.6045251752708731, + "grad_norm": 208.56423950195312, + "kl": 0.1650390625, + "learning_rate": 3.954748247291268e-07, + "loss": 0.0066, + "reward": 1.7036913633346558, + "reward_std": 0.13579218089580536, + "rewards/answer_reward": 0.234375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.46931636333465576, + "step": 1897 + }, + { + "completion_length": 97.1875, + "epoch": 0.6048438495857233, + "grad_norm": 21.571367263793945, + "kl": 0.1484375, + "learning_rate": 3.9515615041427657e-07, + "loss": 0.0059, + "reward": 1.5167163610458374, + "reward_std": 0.10557783395051956, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4073413908481598, + "rewards/pad": 0.109375, + "step": 1898 + }, + { + "completion_length": 45.796875, + "epoch": 0.6051625239005736, + "grad_norm": 17.291685104370117, + "kl": 0.1533203125, + "learning_rate": 3.948374760994264e-07, + "loss": 0.0061, + "reward": 1.7278497219085693, + "reward_std": 0.10903044790029526, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.43097472190856934, + "rewards/pad": 0.296875, + "step": 1899 + }, + { + "completion_length": 93.859375, + "epoch": 0.6054811982154238, + "grad_norm": 23.579120635986328, + "kl": 0.11376953125, + "learning_rate": 3.9451880178457613e-07, + "loss": 0.0046, + "reward": 1.507385015487671, + "reward_std": 0.03943081945180893, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5073848962783813, + "rewards/pad": 0.0, + "step": 1900 + }, + { + "completion_length": 120.734375, + "epoch": 0.605799872530274, + "grad_norm": 100.17448425292969, + "kl": 0.09130859375, + "learning_rate": 3.9420012746972594e-07, + "loss": 0.0037, + "reward": 1.5422768592834473, + "reward_std": 0.13402345776557922, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.43290185928344727, + "rewards/pad": 0.125, + "step": 1901 + }, + { + "completion_length": 45.484375, + "epoch": 0.6061185468451242, + "grad_norm": 29.960926055908203, + "kl": 0.2216796875, + "learning_rate": 3.938814531548757e-07, + "loss": 0.0089, + "reward": 1.6506834030151367, + "reward_std": 0.08238563686609268, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5256834030151367, + "rewards/pad": 0.125, + "step": 1902 + }, + { + "completion_length": 46.140625, + "epoch": 0.6064372211599746, + "grad_norm": 43.66762924194336, + "kl": 0.20703125, + "learning_rate": 3.935627788400255e-07, + "loss": 0.0083, + "reward": 1.7490934133529663, + "reward_std": 0.1362352967262268, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5459684729576111, + "rewards/pad": 0.203125, + "step": 1903 + }, + { + "completion_length": 146.453125, + "epoch": 0.6067558954748248, + "grad_norm": 7.266249656677246, + "kl": 0.10205078125, + "learning_rate": 3.9324410452517525e-07, + "loss": 0.0041, + "reward": 1.4332079887390137, + "reward_std": 0.03245477378368378, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.43320807814598083, + "rewards/pad": 0.0, + "step": 1904 + }, + { + "completion_length": 99.109375, + "epoch": 0.607074569789675, + "grad_norm": 14.11192512512207, + "kl": 0.1630859375, + "learning_rate": 3.9292543021032506e-07, + "loss": 0.0065, + "reward": 1.5477190017700195, + "reward_std": 0.11120744049549103, + "rewards/pad": 0.359375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.18834398686885834, + "step": 1905 + }, + { + "completion_length": 73.515625, + "epoch": 0.6073932441045252, + "grad_norm": 15.46536922454834, + "kl": 0.10791015625, + "learning_rate": 3.926067558954748e-07, + "loss": 0.0043, + "reward": 1.7275207042694092, + "reward_std": 0.0868614986538887, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3525207042694092, + "rewards/pad": 0.375, + "step": 1906 + }, + { + "completion_length": 47.078125, + "epoch": 0.6077119184193754, + "grad_norm": 63.19918441772461, + "kl": 0.171875, + "learning_rate": 3.922880815806246e-07, + "loss": 0.0069, + "reward": 1.7599387168884277, + "reward_std": 0.17778722941875458, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5568135976791382, + "rewards/pad": 0.203125, + "step": 1907 + }, + { + "completion_length": 96.875, + "epoch": 0.6080305927342257, + "grad_norm": 21.277603149414062, + "kl": 0.1337890625, + "learning_rate": 3.919694072657744e-07, + "loss": 0.0054, + "reward": 1.4872418642044067, + "reward_std": 0.0728248655796051, + "rewards/answer_reward": 0.0, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.48724186420440674, + "step": 1908 + }, + { + "completion_length": 98.390625, + "epoch": 0.6083492670490759, + "grad_norm": 30.763784408569336, + "kl": 0.12890625, + "learning_rate": 3.9165073295092413e-07, + "loss": 0.0051, + "reward": 1.6191275119781494, + "reward_std": 0.154721200466156, + "rewards/answer_reward": 0.21875, + "rewards/format_reward_gqa": 0.984375, + "rewards/iou_glue_reward": 0.4160025119781494, + "step": 1909 + }, + { + "completion_length": 70.1875, + "epoch": 0.6086679413639261, + "grad_norm": 26.556743621826172, + "kl": 0.173828125, + "learning_rate": 3.9133205863607394e-07, + "loss": 0.007, + "reward": 1.53456711769104, + "reward_std": 0.06957153975963593, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.53456711769104, + "rewards/pad": 0.0, + "step": 1910 + }, + { + "completion_length": 96.671875, + "epoch": 0.6089866156787763, + "grad_norm": 117.11419677734375, + "kl": 0.1240234375, + "learning_rate": 3.910133843212237e-07, + "loss": 0.005, + "reward": 1.5257726907730103, + "reward_std": 0.04365560784935951, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5257725715637207, + "step": 1911 + }, + { + "completion_length": 121.6875, + "epoch": 0.6093052899936265, + "grad_norm": 31.417213439941406, + "kl": 0.11767578125, + "learning_rate": 3.906947100063735e-07, + "loss": 0.0047, + "reward": 1.5414679050445557, + "reward_std": 0.13280373811721802, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.4320930540561676, + "step": 1912 + }, + { + "completion_length": 97.671875, + "epoch": 0.6096239643084768, + "grad_norm": 61.08513641357422, + "kl": 0.12255859375, + "learning_rate": 3.903760356915232e-07, + "loss": 0.0049, + "reward": 1.5116188526153564, + "reward_std": 0.11132874339818954, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.527243971824646, + "rewards/pad": 0.0, + "step": 1913 + }, + { + "completion_length": 71.25, + "epoch": 0.609942638623327, + "grad_norm": 20.120527267456055, + "kl": 0.2451171875, + "learning_rate": 3.90057361376673e-07, + "loss": 0.0098, + "reward": 1.551047921180725, + "reward_std": 0.07995973527431488, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5510479807853699, + "rewards/pad": 0.0, + "step": 1914 + }, + { + "completion_length": 95.0625, + "epoch": 0.6102613129381772, + "grad_norm": 20.293169021606445, + "kl": 0.404296875, + "learning_rate": 3.8973868706182276e-07, + "loss": 0.0161, + "reward": 1.6185961961746216, + "reward_std": 0.10013433545827866, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.6185961365699768, + "step": 1915 + }, + { + "completion_length": 95.453125, + "epoch": 0.6105799872530274, + "grad_norm": 23.486623764038086, + "kl": 0.10400390625, + "learning_rate": 3.8942001274697257e-07, + "loss": 0.0042, + "reward": 1.4344849586486816, + "reward_std": 0.10762172937393188, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.43448498845100403, + "rewards/pad": 0.0, + "step": 1916 + }, + { + "completion_length": 69.96875, + "epoch": 0.6108986615678776, + "grad_norm": 17.362545013427734, + "kl": 0.244140625, + "learning_rate": 3.891013384321223e-07, + "loss": 0.0098, + "reward": 1.6059834957122803, + "reward_std": 0.11794085055589676, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4966084659099579, + "rewards/pad": 0.109375, + "step": 1917 + }, + { + "completion_length": 123.125, + "epoch": 0.6112173358827279, + "grad_norm": 22.448604583740234, + "kl": 0.12109375, + "learning_rate": 3.8878266411727213e-07, + "loss": 0.0048, + "reward": 1.3911631107330322, + "reward_std": 0.04575665667653084, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3911631107330322, + "step": 1918 + }, + { + "completion_length": 146.09375, + "epoch": 0.6115360101975781, + "grad_norm": 18.462223052978516, + "kl": 0.08447265625, + "learning_rate": 3.884639898024219e-07, + "loss": 0.0034, + "reward": 1.53928542137146, + "reward_std": 0.04528950899839401, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.53928542137146, + "step": 1919 + }, + { + "completion_length": 72.078125, + "epoch": 0.6118546845124283, + "grad_norm": 22.600019454956055, + "kl": 0.1318359375, + "learning_rate": 3.881453154875717e-07, + "loss": 0.0053, + "reward": 1.585951805114746, + "reward_std": 0.12895137071609497, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.6015768647193909, + "step": 1920 + }, + { + "completion_length": 123.5, + "epoch": 0.6121733588272785, + "grad_norm": 17.322956085205078, + "kl": 0.1640625, + "learning_rate": 3.8782664117272144e-07, + "loss": 0.0066, + "reward": 1.679431438446045, + "reward_std": 0.10875079780817032, + "rewards/answer_reward": 0.15625, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5231814384460449, + "step": 1921 + }, + { + "completion_length": 96.921875, + "epoch": 0.6124920331421287, + "grad_norm": 13.455272674560547, + "kl": 0.138671875, + "learning_rate": 3.8750796685787125e-07, + "loss": 0.0055, + "reward": 1.5285980701446533, + "reward_std": 0.06912402808666229, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.40359804034233093, + "rewards/pad": 0.125, + "step": 1922 + }, + { + "completion_length": 70.796875, + "epoch": 0.612810707456979, + "grad_norm": 16.77520179748535, + "kl": 0.263671875, + "learning_rate": 3.87189292543021e-07, + "loss": 0.0106, + "reward": 1.5207405090332031, + "reward_std": 0.07428693771362305, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5207405686378479, + "step": 1923 + }, + { + "completion_length": 45.15625, + "epoch": 0.6131293817718292, + "grad_norm": 72.12005615234375, + "kl": 0.1337890625, + "learning_rate": 3.868706182281708e-07, + "loss": 0.0053, + "reward": 1.4954227209091187, + "reward_std": 0.08774071931838989, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.49542272090911865, + "rewards/pad": 0.0, + "step": 1924 + }, + { + "completion_length": 119.359375, + "epoch": 0.6134480560866794, + "grad_norm": 45.66981506347656, + "kl": 0.0908203125, + "learning_rate": 3.8655194391332057e-07, + "loss": 0.0036, + "reward": 1.488682508468628, + "reward_std": 0.05444840341806412, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4886825680732727, + "rewards/pad": 0.0, + "step": 1925 + }, + { + "completion_length": 70.65625, + "epoch": 0.6137667304015296, + "grad_norm": 43.318458557128906, + "kl": 0.142578125, + "learning_rate": 3.8623326959847037e-07, + "loss": 0.0057, + "reward": 1.457085371017456, + "reward_std": 0.07869704812765121, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.45708543062210083, + "rewards/pad": 0.0, + "step": 1926 + }, + { + "completion_length": 101.078125, + "epoch": 0.6140854047163798, + "grad_norm": 50.392730712890625, + "kl": 0.1103515625, + "learning_rate": 3.8591459528362013e-07, + "loss": 0.0044, + "reward": 1.7335388660430908, + "reward_std": 0.1704206019639969, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4366638660430908, + "rewards/pad": 0.296875, + "step": 1927 + }, + { + "completion_length": 73.296875, + "epoch": 0.6144040790312301, + "grad_norm": 44.92176818847656, + "kl": 0.12109375, + "learning_rate": 3.8559592096876993e-07, + "loss": 0.0048, + "reward": 1.5867115259170532, + "reward_std": 0.18084655702114105, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.446086585521698, + "rewards/pad": 0.140625, + "step": 1928 + }, + { + "completion_length": 123.96875, + "epoch": 0.6147227533460803, + "grad_norm": 11.913430213928223, + "kl": 0.154296875, + "learning_rate": 3.852772466539197e-07, + "loss": 0.0062, + "reward": 1.5406157970428467, + "reward_std": 0.10338909924030304, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5406157970428467, + "step": 1929 + }, + { + "completion_length": 72.4375, + "epoch": 0.6150414276609305, + "grad_norm": 16.6257381439209, + "kl": 0.1630859375, + "learning_rate": 3.849585723390695e-07, + "loss": 0.0065, + "reward": 1.632989525794983, + "reward_std": 0.10083861649036407, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6329895257949829, + "rewards/pad": 0.0, + "step": 1930 + }, + { + "completion_length": 95.296875, + "epoch": 0.6153601019757807, + "grad_norm": 109.17361450195312, + "kl": 0.1376953125, + "learning_rate": 3.8463989802421925e-07, + "loss": 0.0055, + "reward": 1.5289793014526367, + "reward_std": 0.09404543787240982, + "rewards/answer_reward": 0.03125, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4977293312549591, + "step": 1931 + }, + { + "completion_length": 123.578125, + "epoch": 0.615678776290631, + "grad_norm": 16.52717399597168, + "kl": 0.2021484375, + "learning_rate": 3.84321223709369e-07, + "loss": 0.0081, + "reward": 1.5912632942199707, + "reward_std": 0.07738955318927765, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5912632346153259, + "step": 1932 + }, + { + "completion_length": 122.984375, + "epoch": 0.6159974506054812, + "grad_norm": 22.1501522064209, + "kl": 0.1318359375, + "learning_rate": 3.8400254939451876e-07, + "loss": 0.0053, + "reward": 1.4621679782867432, + "reward_std": 0.07366758584976196, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.46216797828674316, + "step": 1933 + }, + { + "completion_length": 97.1875, + "epoch": 0.6163161249203314, + "grad_norm": 21.725025177001953, + "kl": 0.1064453125, + "learning_rate": 3.8368387507966856e-07, + "loss": 0.0043, + "reward": 1.3546940088272095, + "reward_std": 0.06960678845643997, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3546939492225647, + "step": 1934 + }, + { + "completion_length": 71.34375, + "epoch": 0.6166347992351816, + "grad_norm": 24.456161499023438, + "kl": 0.1630859375, + "learning_rate": 3.833652007648183e-07, + "loss": 0.0065, + "reward": 1.5506436824798584, + "reward_std": 0.09592757374048233, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4256436228752136, + "step": 1935 + }, + { + "completion_length": 72.5, + "epoch": 0.6169534735500318, + "grad_norm": 20.02235221862793, + "kl": 0.146484375, + "learning_rate": 3.830465264499681e-07, + "loss": 0.0059, + "reward": 1.6763842105865479, + "reward_std": 0.05954057723283768, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5513842105865479, + "rewards/pad": 0.125, + "step": 1936 + }, + { + "completion_length": 97.84375, + "epoch": 0.617272147864882, + "grad_norm": 116.22586059570312, + "kl": 0.125, + "learning_rate": 3.827278521351179e-07, + "loss": 0.005, + "reward": 1.5901174545288086, + "reward_std": 0.050210874527692795, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5901174545288086, + "step": 1937 + }, + { + "completion_length": 99.4375, + "epoch": 0.6175908221797323, + "grad_norm": 33.92824935913086, + "kl": 0.1708984375, + "learning_rate": 3.8240917782026763e-07, + "loss": 0.0068, + "reward": 1.612595796585083, + "reward_std": 0.053627826273441315, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.48759573698043823, + "rewards/pad": 0.125, + "step": 1938 + }, + { + "completion_length": 123.0625, + "epoch": 0.6179094964945825, + "grad_norm": 21.3590087890625, + "kl": 0.0830078125, + "learning_rate": 3.8209050350541744e-07, + "loss": 0.0033, + "reward": 1.4808800220489502, + "reward_std": 0.07551354914903641, + "rewards/pad": 0.015625, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4652549922466278, + "step": 1939 + }, + { + "completion_length": 121.28125, + "epoch": 0.6182281708094327, + "grad_norm": 31.41773223876953, + "kl": 0.166015625, + "learning_rate": 3.817718291905672e-07, + "loss": 0.0066, + "reward": 1.5100470781326294, + "reward_std": 0.08721407502889633, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5100470781326294, + "step": 1940 + }, + { + "completion_length": 45.234375, + "epoch": 0.6185468451242829, + "grad_norm": 104.99120330810547, + "kl": 0.140625, + "learning_rate": 3.81453154875717e-07, + "loss": 0.0056, + "reward": 1.445270299911499, + "reward_std": 0.1404525339603424, + "rewards/answer_reward": 0.0, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.4452703297138214, + "step": 1941 + }, + { + "completion_length": 100.296875, + "epoch": 0.6188655194391333, + "grad_norm": 37.74537658691406, + "kl": 0.2314453125, + "learning_rate": 3.8113448056086675e-07, + "loss": 0.0092, + "reward": 1.700887680053711, + "reward_std": 0.08685436844825745, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4508877396583557, + "rewards/pad": 0.25, + "step": 1942 + }, + { + "completion_length": 18.375, + "epoch": 0.6191841937539835, + "grad_norm": 22.813129425048828, + "kl": 0.2216796875, + "learning_rate": 3.8081580624601656e-07, + "loss": 0.0088, + "reward": 1.7529795169830322, + "reward_std": 0.08033765107393265, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.7529796361923218, + "rewards/pad": 0.0, + "step": 1943 + }, + { + "completion_length": 148.234375, + "epoch": 0.6195028680688337, + "grad_norm": 93.70587921142578, + "kl": 0.0859375, + "learning_rate": 3.804971319311663e-07, + "loss": 0.0034, + "reward": 1.4083514213562012, + "reward_std": 0.042352207005023956, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.408351331949234, + "rewards/pad": 0.0, + "step": 1944 + }, + { + "completion_length": 73.3125, + "epoch": 0.6198215423836839, + "grad_norm": 57.26400375366211, + "kl": 0.1611328125, + "learning_rate": 3.801784576163161e-07, + "loss": 0.0065, + "reward": 1.7585382461547852, + "reward_std": 0.06438243389129639, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5085383057594299, + "step": 1945 + }, + { + "completion_length": 146.8125, + "epoch": 0.6201402166985341, + "grad_norm": 10.125571250915527, + "kl": 0.11083984375, + "learning_rate": 3.798597833014659e-07, + "loss": 0.0044, + "reward": 1.5111711025238037, + "reward_std": 0.03506201505661011, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5111711025238037, + "step": 1946 + }, + { + "completion_length": 73.359375, + "epoch": 0.6204588910133844, + "grad_norm": 14.61278247833252, + "kl": 0.3203125, + "learning_rate": 3.795411089866157e-07, + "loss": 0.0128, + "reward": 1.724852442741394, + "reward_std": 0.19486337900161743, + "rewards/answer_reward": 0.21875, + "rewards/format_reward_gqa": 0.984375, + "rewards/iou_glue_reward": 0.521727442741394, + "step": 1947 + }, + { + "completion_length": 123.0, + "epoch": 0.6207775653282346, + "grad_norm": 19.849451065063477, + "kl": 0.10986328125, + "learning_rate": 3.7922243467176544e-07, + "loss": 0.0044, + "reward": 1.784373164176941, + "reward_std": 0.062317900359630585, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5343732237815857, + "rewards/pad": 0.25, + "step": 1948 + }, + { + "completion_length": 97.890625, + "epoch": 0.6210962396430848, + "grad_norm": 18.029869079589844, + "kl": 0.1279296875, + "learning_rate": 3.7890376035691524e-07, + "loss": 0.0051, + "reward": 1.6952381134033203, + "reward_std": 0.06585781276226044, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5702381730079651, + "step": 1949 + }, + { + "completion_length": 128.25, + "epoch": 0.621414913957935, + "grad_norm": 80.95796966552734, + "kl": 0.09912109375, + "learning_rate": 3.78585086042065e-07, + "loss": 0.004, + "reward": 1.4363656044006348, + "reward_std": 0.04811306297779083, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.43636554479599, + "rewards/pad": 0.0, + "step": 1950 + }, + { + "completion_length": 98.75, + "epoch": 0.6217335882727852, + "grad_norm": 52.44783020019531, + "kl": 0.1572265625, + "learning_rate": 3.782664117272148e-07, + "loss": 0.0063, + "reward": 1.5792521238327026, + "reward_std": 0.10398217290639877, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.45425212383270264, + "rewards/pad": 0.125, + "step": 1951 + }, + { + "completion_length": 71.953125, + "epoch": 0.6220522625876355, + "grad_norm": 17.435596466064453, + "kl": 0.11572265625, + "learning_rate": 3.779477374123645e-07, + "loss": 0.0046, + "reward": 1.7964942455291748, + "reward_std": 0.05092811957001686, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6714942455291748, + "rewards/pad": 0.125, + "step": 1952 + }, + { + "completion_length": 72.078125, + "epoch": 0.6223709369024857, + "grad_norm": 27.534250259399414, + "kl": 0.119140625, + "learning_rate": 3.776290630975143e-07, + "loss": 0.0048, + "reward": 1.6906150579452515, + "reward_std": 0.03630218282341957, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.565615177154541, + "step": 1953 + }, + { + "completion_length": 99.640625, + "epoch": 0.6226896112173359, + "grad_norm": 42.14917755126953, + "kl": 0.10546875, + "learning_rate": 3.7731038878266407e-07, + "loss": 0.0042, + "reward": 1.6329374313354492, + "reward_std": 0.1385832130908966, + "rewards/pad": 0.015625, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.632937490940094, + "step": 1954 + }, + { + "completion_length": 71.421875, + "epoch": 0.6230082855321861, + "grad_norm": 76.2291030883789, + "kl": 0.1748046875, + "learning_rate": 3.769917144678139e-07, + "loss": 0.007, + "reward": 1.5288151502609253, + "reward_std": 0.12570810317993164, + "rewards/pad": 0.140625, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.3881901800632477, + "step": 1955 + }, + { + "completion_length": 69.546875, + "epoch": 0.6233269598470363, + "grad_norm": 40.875633239746094, + "kl": 0.1767578125, + "learning_rate": 3.7667304015296363e-07, + "loss": 0.0071, + "reward": 1.4328603744506836, + "reward_std": 0.10027036815881729, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.432860404253006, + "step": 1956 + }, + { + "completion_length": 101.28125, + "epoch": 0.6236456341618866, + "grad_norm": 27.414457321166992, + "kl": 0.27734375, + "learning_rate": 3.7635436583811344e-07, + "loss": 0.0111, + "reward": 1.7710248231887817, + "reward_std": 0.14013846218585968, + "rewards/pad": 0.359375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4116498827934265, + "step": 1957 + }, + { + "completion_length": 44.296875, + "epoch": 0.6239643084767368, + "grad_norm": 30.38433074951172, + "kl": 0.2255859375, + "learning_rate": 3.760356915232632e-07, + "loss": 0.009, + "reward": 1.5368155241012573, + "reward_std": 0.09000550210475922, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5368155241012573, + "rewards/pad": 0.0, + "step": 1958 + }, + { + "completion_length": 71.265625, + "epoch": 0.624282982791587, + "grad_norm": 73.53339385986328, + "kl": 0.169921875, + "learning_rate": 3.75717017208413e-07, + "loss": 0.0068, + "reward": 1.6425340175628662, + "reward_std": 0.14571252465248108, + "rewards/format_reward_tg": 0.96875, + "rewards/iou_timestamp_reward": 0.564409077167511, + "rewards/pad": 0.109375, + "step": 1959 + }, + { + "completion_length": 97.953125, + "epoch": 0.6246016571064372, + "grad_norm": 63.19132995605469, + "kl": 1.2578125, + "learning_rate": 3.7539834289356275e-07, + "loss": 0.05, + "reward": 1.7158057689666748, + "reward_std": 0.10019126534461975, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.48143070936203003, + "rewards/pad": 0.234375, + "step": 1960 + }, + { + "completion_length": 150.265625, + "epoch": 0.6249203314212874, + "grad_norm": 54.07538986206055, + "kl": 0.09716796875, + "learning_rate": 3.7507966857871256e-07, + "loss": 0.0039, + "reward": 1.510149598121643, + "reward_std": 0.08005047589540482, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3851495385169983, + "rewards/pad": 0.125, + "step": 1961 + }, + { + "completion_length": 122.15625, + "epoch": 0.6252390057361377, + "grad_norm": 21.607746124267578, + "kl": 0.109375, + "learning_rate": 3.747609942638623e-07, + "loss": 0.0044, + "reward": 1.5534250736236572, + "reward_std": 0.04033489525318146, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5534250736236572, + "step": 1962 + }, + { + "completion_length": 70.15625, + "epoch": 0.6255576800509879, + "grad_norm": 15.613412857055664, + "kl": 0.1416015625, + "learning_rate": 3.744423199490121e-07, + "loss": 0.0056, + "reward": 1.618166208267212, + "reward_std": 0.16496075689792633, + "rewards/pad": 0.078125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5400412082672119, + "step": 1963 + }, + { + "completion_length": 98.703125, + "epoch": 0.6258763543658381, + "grad_norm": 37.2735481262207, + "kl": 0.1796875, + "learning_rate": 3.7412364563416187e-07, + "loss": 0.0072, + "reward": 1.6815367937088013, + "reward_std": 0.09005028754472733, + "rewards/pad": 0.140625, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5409117937088013, + "step": 1964 + }, + { + "completion_length": 70.953125, + "epoch": 0.6261950286806883, + "grad_norm": 23.696840286254883, + "kl": 0.1337890625, + "learning_rate": 3.738049713193117e-07, + "loss": 0.0053, + "reward": 1.6503503322601318, + "reward_std": 0.05913899093866348, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5253504514694214, + "step": 1965 + }, + { + "completion_length": 123.890625, + "epoch": 0.6265137029955385, + "grad_norm": 31.705644607543945, + "kl": 0.12890625, + "learning_rate": 3.7348629700446143e-07, + "loss": 0.0051, + "reward": 1.4367663860321045, + "reward_std": 0.07267533242702484, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4367663264274597, + "step": 1966 + }, + { + "completion_length": 73.40625, + "epoch": 0.6268323773103888, + "grad_norm": 140.63800048828125, + "kl": 0.185546875, + "learning_rate": 3.7316762268961124e-07, + "loss": 0.0074, + "reward": 1.666994333267212, + "reward_std": 0.08498790860176086, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6669944524765015, + "rewards/pad": 0.0, + "step": 1967 + }, + { + "completion_length": 72.296875, + "epoch": 0.627151051625239, + "grad_norm": 35.06744384765625, + "kl": 0.1396484375, + "learning_rate": 3.72848948374761e-07, + "loss": 0.0056, + "reward": 1.6456339359283447, + "reward_std": 0.0863327831029892, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5206338167190552, + "rewards/pad": 0.125, + "step": 1968 + }, + { + "completion_length": 124.296875, + "epoch": 0.6274697259400892, + "grad_norm": 46.038536071777344, + "kl": 0.099609375, + "learning_rate": 3.725302740599108e-07, + "loss": 0.004, + "reward": 1.6347662210464478, + "reward_std": 0.08927043527364731, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5097662806510925, + "rewards/pad": 0.125, + "step": 1969 + }, + { + "completion_length": 96.609375, + "epoch": 0.6277884002549394, + "grad_norm": 43.614681243896484, + "kl": 0.1337890625, + "learning_rate": 3.7221159974506056e-07, + "loss": 0.0053, + "reward": 1.7559230327606201, + "reward_std": 0.15201693773269653, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5840479731559753, + "rewards/pad": 0.171875, + "step": 1970 + }, + { + "completion_length": 43.640625, + "epoch": 0.6281070745697896, + "grad_norm": 175.77879333496094, + "kl": 0.34375, + "learning_rate": 3.718929254302103e-07, + "loss": 0.0137, + "reward": 1.989320993423462, + "reward_std": 0.11485286056995392, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.7393209934234619, + "rewards/pad": 0.25, + "step": 1971 + }, + { + "completion_length": 148.3125, + "epoch": 0.6284257488846399, + "grad_norm": 30.099884033203125, + "kl": 0.0986328125, + "learning_rate": 3.7157425111536006e-07, + "loss": 0.0039, + "reward": 1.3178768157958984, + "reward_std": 0.09088070690631866, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.33350178599357605, + "step": 1972 + }, + { + "completion_length": 73.5625, + "epoch": 0.6287444231994901, + "grad_norm": 13.46129035949707, + "kl": 0.130859375, + "learning_rate": 3.712555768005098e-07, + "loss": 0.0052, + "reward": 1.5966365337371826, + "reward_std": 0.1090199202299118, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.2841365933418274, + "rewards/pad": 0.3125, + "step": 1973 + }, + { + "completion_length": 70.40625, + "epoch": 0.6290630975143403, + "grad_norm": 20.175935745239258, + "kl": 0.20703125, + "learning_rate": 3.709369024856596e-07, + "loss": 0.0083, + "reward": 1.5837442874908447, + "reward_std": 0.07582993805408478, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5837444067001343, + "rewards/pad": 0.0, + "step": 1974 + }, + { + "completion_length": 44.59375, + "epoch": 0.6293817718291905, + "grad_norm": 49.37710952758789, + "kl": 0.18359375, + "learning_rate": 3.706182281708094e-07, + "loss": 0.0073, + "reward": 1.6172418594360352, + "reward_std": 0.10630054026842117, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6172418594360352, + "rewards/pad": 0.0, + "step": 1975 + }, + { + "completion_length": 72.390625, + "epoch": 0.6297004461440407, + "grad_norm": 18.700769424438477, + "kl": 0.2021484375, + "learning_rate": 3.702995538559592e-07, + "loss": 0.0081, + "reward": 1.7520803213119507, + "reward_std": 0.05088932067155838, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.5020803213119507, + "step": 1976 + }, + { + "completion_length": 124.703125, + "epoch": 0.630019120458891, + "grad_norm": 70.51777648925781, + "kl": 0.08837890625, + "learning_rate": 3.6998087954110894e-07, + "loss": 0.0035, + "reward": 1.7651245594024658, + "reward_std": 0.0476035512983799, + "rewards/answer_reward": 0.375, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.39012449979782104, + "step": 1977 + }, + { + "completion_length": 43.921875, + "epoch": 0.6303377947737412, + "grad_norm": 59.889183044433594, + "kl": 0.1748046875, + "learning_rate": 3.6966220522625875e-07, + "loss": 0.007, + "reward": 1.6480891704559326, + "reward_std": 0.12159959971904755, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.6480890512466431, + "step": 1978 + }, + { + "completion_length": 74.578125, + "epoch": 0.6306564690885914, + "grad_norm": 23.10746955871582, + "kl": 0.13671875, + "learning_rate": 3.693435309114085e-07, + "loss": 0.0055, + "reward": 1.703021764755249, + "reward_std": 0.1205596774816513, + "rewards/pad": 0.171875, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5311467051506042, + "step": 1979 + }, + { + "completion_length": 121.375, + "epoch": 0.6309751434034416, + "grad_norm": 18.383968353271484, + "kl": 0.1982421875, + "learning_rate": 3.690248565965583e-07, + "loss": 0.0079, + "reward": 1.4522733688354492, + "reward_std": 0.11010268330574036, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4522734582424164, + "rewards/pad": 0.0, + "step": 1980 + }, + { + "completion_length": 49.40625, + "epoch": 0.631293817718292, + "grad_norm": 39.16129684448242, + "kl": 0.447265625, + "learning_rate": 3.6870618228170806e-07, + "loss": 0.0179, + "reward": 1.5960018634796143, + "reward_std": 0.22656530141830444, + "rewards/answer_reward": 0.25, + "rewards/format_reward_gqa": 1.0, + "rewards/iou_glue_reward": 0.3460018038749695, + "step": 1981 + }, + { + "completion_length": 97.59375, + "epoch": 0.6316124920331422, + "grad_norm": 56.11304473876953, + "kl": 0.14453125, + "learning_rate": 3.6838750796685787e-07, + "loss": 0.0058, + "reward": 1.5802655220031738, + "reward_std": 0.07598498463630676, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.45526552200317383, + "rewards/pad": 0.125, + "step": 1982 + }, + { + "completion_length": 69.96875, + "epoch": 0.6319311663479924, + "grad_norm": 14.71876335144043, + "kl": 0.2490234375, + "learning_rate": 3.680688336520076e-07, + "loss": 0.01, + "reward": 1.604870319366455, + "reward_std": 0.09864290058612823, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6048702001571655, + "rewards/pad": 0.0, + "step": 1983 + }, + { + "completion_length": 76.15625, + "epoch": 0.6322498406628426, + "grad_norm": 45.49012756347656, + "kl": 0.271484375, + "learning_rate": 3.6775015933715743e-07, + "loss": 0.0109, + "reward": 2.0544075965881348, + "reward_std": 0.10176599770784378, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5544075965881348, + "rewards/pad": 0.5, + "step": 1984 + }, + { + "completion_length": 96.609375, + "epoch": 0.6325685149776928, + "grad_norm": 26.258607864379883, + "kl": 0.1591796875, + "learning_rate": 3.674314850223072e-07, + "loss": 0.0064, + "reward": 1.666446566581726, + "reward_std": 0.12379075586795807, + "rewards/pad": 0.109375, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5570715665817261, + "step": 1985 + }, + { + "completion_length": 122.3125, + "epoch": 0.6328871892925431, + "grad_norm": 35.372032165527344, + "kl": 0.10302734375, + "learning_rate": 3.67112810707457e-07, + "loss": 0.0041, + "reward": 1.5771713256835938, + "reward_std": 0.03920065239071846, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5771713256835938, + "step": 1986 + }, + { + "completion_length": 45.859375, + "epoch": 0.6332058636073933, + "grad_norm": 57.8635139465332, + "kl": 0.2236328125, + "learning_rate": 3.6679413639260674e-07, + "loss": 0.0089, + "reward": 1.5414435863494873, + "reward_std": 0.15431562066078186, + "rewards/format_reward_tg": 0.984375, + "rewards/iou_timestamp_reward": 0.5414435863494873, + "rewards/pad": 0.015625, + "step": 1987 + }, + { + "completion_length": 100.015625, + "epoch": 0.6335245379222435, + "grad_norm": 19.29568099975586, + "kl": 0.1025390625, + "learning_rate": 3.6647546207775655e-07, + "loss": 0.0041, + "reward": 1.4702959060668945, + "reward_std": 0.07151374965906143, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.34529587626457214, + "step": 1988 + }, + { + "completion_length": 97.109375, + "epoch": 0.6338432122370937, + "grad_norm": 19.67873764038086, + "kl": 0.123046875, + "learning_rate": 3.661567877629063e-07, + "loss": 0.0049, + "reward": 1.690169095993042, + "reward_std": 0.043867893517017365, + "rewards/pad": 0.125, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5651690363883972, + "step": 1989 + }, + { + "completion_length": 18.921875, + "epoch": 0.6341618865519439, + "grad_norm": 32.41449737548828, + "kl": 0.173828125, + "learning_rate": 3.658381134480561e-07, + "loss": 0.0069, + "reward": 1.4949960708618164, + "reward_std": 0.11608640849590302, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4949961006641388, + "rewards/pad": 0.0, + "step": 1990 + }, + { + "completion_length": 46.328125, + "epoch": 0.6344805608667942, + "grad_norm": 22.01445198059082, + "kl": 0.26953125, + "learning_rate": 3.655194391332058e-07, + "loss": 0.0108, + "reward": 1.8105196952819824, + "reward_std": 0.08823952078819275, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5605197548866272, + "rewards/pad": 0.25, + "step": 1991 + }, + { + "completion_length": 95.953125, + "epoch": 0.6347992351816444, + "grad_norm": 26.866201400756836, + "kl": 0.169921875, + "learning_rate": 3.652007648183556e-07, + "loss": 0.0068, + "reward": 1.4317470788955688, + "reward_std": 0.0999232605099678, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.4317470192909241, + "step": 1992 + }, + { + "completion_length": 68.734375, + "epoch": 0.6351179094964946, + "grad_norm": 33.86798858642578, + "kl": 0.1513671875, + "learning_rate": 3.648820905035054e-07, + "loss": 0.006, + "reward": 1.461557388305664, + "reward_std": 0.043246183544397354, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.4615574777126312, + "rewards/pad": 0.0, + "step": 1993 + }, + { + "completion_length": 97.0625, + "epoch": 0.6354365838113448, + "grad_norm": 22.541500091552734, + "kl": 0.142578125, + "learning_rate": 3.645634161886552e-07, + "loss": 0.0057, + "reward": 1.4119422435760498, + "reward_std": 0.10992357134819031, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.4275672137737274, + "step": 1994 + }, + { + "completion_length": 71.453125, + "epoch": 0.635755258126195, + "grad_norm": 54.545433044433594, + "kl": 1.0703125, + "learning_rate": 3.6424474187380494e-07, + "loss": 0.0428, + "reward": 1.6790932416915894, + "reward_std": 0.1349036693572998, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.6009682416915894, + "rewards/pad": 0.078125, + "step": 1995 + }, + { + "completion_length": 71.328125, + "epoch": 0.6360739324410453, + "grad_norm": 31.94515609741211, + "kl": 0.2314453125, + "learning_rate": 3.6392606755895474e-07, + "loss": 0.0093, + "reward": 1.527273178100586, + "reward_std": 0.1572173535823822, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5272730588912964, + "rewards/pad": 0.0, + "step": 1996 + }, + { + "completion_length": 123.921875, + "epoch": 0.6363926067558955, + "grad_norm": 44.97288513183594, + "kl": 0.166015625, + "learning_rate": 3.636073932441045e-07, + "loss": 0.0066, + "reward": 1.5373291969299316, + "reward_std": 0.10828669369220734, + "rewards/pad": 0.0, + "rewards/tracking_format_reward": 0.984375, + "rewards/tracking_iou_reward": 0.5529541969299316, + "step": 1997 + }, + { + "completion_length": 124.609375, + "epoch": 0.6367112810707457, + "grad_norm": 41.73975372314453, + "kl": 0.09423828125, + "learning_rate": 3.632887189292543e-07, + "loss": 0.0038, + "reward": 1.7525591850280762, + "reward_std": 0.09953247010707855, + "rewards/pad": 0.21875, + "rewards/tracking_format_reward": 1.0, + "rewards/tracking_iou_reward": 0.5338091850280762, + "step": 1998 + }, + { + "completion_length": 70.5625, + "epoch": 0.6370299553855959, + "grad_norm": 63.7142333984375, + "kl": 0.2314453125, + "learning_rate": 3.6297004461440406e-07, + "loss": 0.0092, + "reward": 1.5226670503616333, + "reward_std": 0.08986333757638931, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.5226670503616333, + "rewards/pad": 0.0, + "step": 1999 + }, + { + "completion_length": 98.171875, + "epoch": 0.6373486297004461, + "grad_norm": 73.62125396728516, + "kl": 0.10693359375, + "learning_rate": 3.626513702995538e-07, + "loss": 0.0043, + "reward": 1.3816452026367188, + "reward_std": 0.06278573721647263, + "rewards/format_reward_tg": 1.0, + "rewards/iou_timestamp_reward": 0.3816451132297516, + "rewards/pad": 0.0, + "step": 2000 + } + ], + "logging_steps": 1.0, + "max_steps": 3138, + "num_input_tokens_seen": 0, + "num_train_epochs": 1, + "save_steps": 1000, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 0.0, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +}