Upload args.json
Browse files
args.json
ADDED
|
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"dataset_name": "c4",
|
| 3 |
+
"max_seq_len": 64,
|
| 4 |
+
"enc_dec_model": "facebook/bart-base",
|
| 5 |
+
"train_batch_size": 64,
|
| 6 |
+
"eval_batch_size": 32,
|
| 7 |
+
"num_encoder_latents": 32,
|
| 8 |
+
"num_decoder_latents": 32,
|
| 9 |
+
"dim_ae": 64,
|
| 10 |
+
"num_layers": 3,
|
| 11 |
+
"l2_normalize_latents": true,
|
| 12 |
+
"output_dir": "saved_latent_models/c4/2024-09-22_02-30-27",
|
| 13 |
+
"save_dir": "saved_latent_models",
|
| 14 |
+
"learning_rate": 0.0001,
|
| 15 |
+
"num_train_steps": 50000,
|
| 16 |
+
"lr_schedule": "linear",
|
| 17 |
+
"lr_warmup_steps": 1000,
|
| 18 |
+
"optimizer": "adamw",
|
| 19 |
+
"adam_beta1": 0.9,
|
| 20 |
+
"adam_beta2": 0.999,
|
| 21 |
+
"adam_weight_decay": 0.01,
|
| 22 |
+
"eval_every": 10000,
|
| 23 |
+
"mixed_precision": "no",
|
| 24 |
+
"wandb_name": "bart-roc-l2norm-test-32-64",
|
| 25 |
+
"lm_mode": "freeze",
|
| 26 |
+
"eval": false,
|
| 27 |
+
"resume_training": false,
|
| 28 |
+
"resume_dir": null,
|
| 29 |
+
"num_devices": 1
|
| 30 |
+
}
|