TR-Fin-Table-Structure-HoixiFinetuned-Overdose

This model is a fine-tuned version of microsoft/table-transformer-structure-recognition-v1.1-all on the tr-fin_table dataset. It achieves the following results on the evaluation set:

• Loss: 1.3453

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 1e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• training_steps: 10000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
2.5699	12.5	50	1.8167
2.4601	25.0	100	1.6425
2.2612	37.5	150	1.6247
2.2122	50.0	200	1.5704
1.8015	62.5	250	1.5473
2.0555	75.0	300	1.5278
2.3249	87.5	350	1.5308
1.4308	100.0	400	1.5316
1.9009	112.5	450	1.5131
2.2087	125.0	500	1.5006
2.0473	137.5	550	1.5212
2.5294	150.0	600	1.5379
1.9833	162.5	650	1.5652
1.8314	175.0	700	1.4976
2.4367	187.5	750	1.5416
1.9989	200.0	800	1.5043
1.8754	212.5	850	1.5266
2.0064	225.0	900	1.5051
2.0185	237.5	950	1.5313
1.6792	250.0	1000	1.4967
1.7105	262.5	1050	1.4732
1.9245	275.0	1100	1.5032
1.4862	287.5	1150	1.5018
2.1234	300.0	1200	1.4667
1.7904	312.5	1250	1.4761
2.4147	325.0	1300	1.4684
1.8115	337.5	1350	1.4711
1.9528	350.0	1400	1.4457
1.9951	362.5	1450	1.4618
1.5924	375.0	1500	1.4597
1.9166	387.5	1550	1.4513
1.8729	400.0	1600	1.4332
2.0891	412.5	1650	1.4299
2.1724	425.0	1700	1.4578
2.0618	437.5	1750	1.4791
1.7166	450.0	1800	1.4734
2.027	462.5	1850	1.4736
1.7804	475.0	1900	1.4461
2.3921	487.5	1950	1.4398
1.8792	500.0	2000	1.4345
2.1287	512.5	2050	1.4477
1.5598	525.0	2100	1.4608
1.7381	537.5	2150	1.4428
1.8059	550.0	2200	1.4423
1.7971	562.5	2250	1.4109
1.6301	575.0	2300	1.4341
1.9655	587.5	2350	1.4502
1.4477	600.0	2400	1.4378
1.7368	612.5	2450	1.4445
1.9277	625.0	2500	1.4349
1.8093	637.5	2550	1.4542
1.8594	650.0	2600	1.4462
1.7637	662.5	2650	1.4250
1.9192	675.0	2700	1.4681
1.86	687.5	2750	1.4827
1.8954	700.0	2800	1.4187
1.4728	712.5	2850	1.4129
1.6828	725.0	2900	1.4113
1.7694	737.5	2950	1.4035
1.805	750.0	3000	1.4134
1.8506	762.5	3050	1.4135
1.8127	775.0	3100	1.4057
1.9829	787.5	3150	1.4102
2.0491	800.0	3200	1.4216
1.549	812.5	3250	1.3648
1.8095	825.0	3300	1.4064
1.6556	837.5	3350	1.3776
1.5418	850.0	3400	1.3942
1.5569	862.5	3450	1.4009
2.3074	875.0	3500	1.4011
1.6733	887.5	3550	1.4032
1.9263	900.0	3600	1.3799
1.9946	912.5	3650	1.3972
1.4845	925.0	3700	1.3853
1.9587	937.5	3750	1.4146
1.7828	950.0	3800	1.4037
2.012	962.5	3850	1.4235
1.817	975.0	3900	1.4007
2.0893	987.5	3950	1.4095
2.1338	1000.0	4000	1.3824
1.8228	1012.5	4050	1.3843
1.6272	1025.0	4100	1.4122
1.6202	1037.5	4150	1.3909
1.4482	1050.0	4200	1.3589
1.949	1062.5	4250	1.3605
2.0954	1075.0	4300	1.3869
1.4728	1087.5	4350	1.3944
1.5916	1100.0	4400	1.3825
1.7988	1112.5	4450	1.3682
1.5051	1125.0	4500	1.3719
1.8492	1137.5	4550	1.4000
1.6146	1150.0	4600	1.3886
1.9732	1162.5	4650	1.3769
1.7256	1175.0	4700	1.3717
1.9683	1187.5	4750	1.3849
1.6818	1200.0	4800	1.3951
1.5879	1212.5	4850	1.3903
1.8743	1225.0	4900	1.3988
1.7887	1237.5	4950	1.3970
1.7302	1250.0	5000	1.3774
1.6503	1262.5	5050	1.4183
1.6207	1275.0	5100	1.3892
1.9589	1287.5	5150	1.4226
1.9163	1300.0	5200	1.4142
1.869	1312.5	5250	1.3777
1.601	1325.0	5300	1.3743
1.5548	1337.5	5350	1.3871
1.6482	1350.0	5400	1.4068
1.545	1362.5	5450	1.4012
1.292	1375.0	5500	1.4138
1.5313	1387.5	5550	1.4066
1.5981	1400.0	5600	1.4022
1.6622	1412.5	5650	1.4069
1.7446	1425.0	5700	1.3957
1.9459	1437.5	5750	1.4085
1.6468	1450.0	5800	1.4191
1.6107	1462.5	5850	1.3973
1.5986	1475.0	5900	1.3834
1.6157	1487.5	5950	1.3983
1.7203	1500.0	6000	1.3696
1.7985	1512.5	6050	1.3884
1.9865	1525.0	6100	1.3951
1.5754	1537.5	6150	1.3935
1.7058	1550.0	6200	1.3856
1.7909	1562.5	6250	1.3916
2.0516	1575.0	6300	1.3532
1.787	1587.5	6350	1.4099
1.6804	1600.0	6400	1.4122
1.8824	1612.5	6450	1.3876
1.4672	1625.0	6500	1.3845
1.5871	1637.5	6550	1.3900
1.899	1650.0	6600	1.3777
1.3322	1662.5	6650	1.3765
1.6055	1675.0	6700	1.3556
2.226	1687.5	6750	1.3798
1.3981	1700.0	6800	1.3695
1.6295	1712.5	6850	1.3579
1.5333	1725.0	6900	1.3714
1.5442	1737.5	6950	1.3709
1.2871	1750.0	7000	1.3615
1.6814	1762.5	7050	1.3742
1.4199	1775.0	7100	1.3683
1.6349	1787.5	7150	1.3593
1.4781	1800.0	7200	1.3633
1.9904	1812.5	7250	1.3705
1.6171	1825.0	7300	1.3768
1.7736	1837.5	7350	1.3753
1.7629	1850.0	7400	1.3719
1.6829	1862.5	7450	1.3687
1.4467	1875.0	7500	1.3606
1.8322	1887.5	7550	1.3759
1.9977	1900.0	7600	1.3839
1.6281	1912.5	7650	1.3877
1.4727	1925.0	7700	1.3922
1.739	1937.5	7750	1.3922
2.0781	1950.0	7800	1.4001
1.8195	1962.5	7850	1.3875
1.7775	1975.0	7900	1.3743
1.5131	1987.5	7950	1.3774
1.5687	2000.0	8000	1.3767
1.6019	2012.5	8050	1.3773
1.2421	2025.0	8100	1.3663
1.5391	2037.5	8150	1.3599
1.8665	2050.0	8200	1.3744
1.7484	2062.5	8250	1.3667
1.5384	2075.0	8300	1.3483
1.4885	2087.5	8350	1.3664
1.8017	2100.0	8400	1.3662
1.4904	2112.5	8450	1.3577
1.6576	2125.0	8500	1.3727
1.5057	2137.5	8550	1.3647
1.8728	2150.0	8600	1.3558
1.8287	2162.5	8650	1.3604
1.4705	2175.0	8700	1.3586
1.6126	2187.5	8750	1.3818
1.6838	2200.0	8800	1.3756
1.5985	2212.5	8850	1.3683
1.9316	2225.0	8900	1.3554
1.7605	2237.5	8950	1.3485
1.8473	2250.0	9000	1.3679
1.5161	2262.5	9050	1.3440
1.38	2275.0	9100	1.3578
1.2987	2287.5	9150	1.3477
1.6364	2300.0	9200	1.3497
1.3951	2312.5	9250	1.3630
1.3344	2325.0	9300	1.3498
1.3916	2337.5	9350	1.3503
1.7832	2350.0	9400	1.3502
1.377	2362.5	9450	1.3512
1.3797	2375.0	9500	1.3507
1.4729	2387.5	9550	1.3533
1.5299	2400.0	9600	1.3544
1.6858	2412.5	9650	1.3447
1.3794	2425.0	9700	1.3432
1.8406	2437.5	9750	1.3449
1.8643	2450.0	9800	1.3394
1.5886	2462.5	9850	1.3452
2.065	2475.0	9900	1.3461
1.7918	2487.5	9950	1.3447
1.3398	2500.0	10000	1.3453

Framework versions

• Transformers 4.51.3
• Pytorch 2.6.0+cu126
• Datasets 3.5.0
• Tokenizers 0.21.0