table-transformer-structure-recognition-v1.1-all-finetuned-v4

This model is a fine-tuned version of microsoft/table-transformer-structure-recognition-v1.1-all on the tr-fin_table-dataset-v4 dataset. It achieves the following results on the evaluation set:

• Loss: 1.5814

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: 5e-06
• 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: 1000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
2.8161	12.5	50	1.8594
2.1356	25.0	100	1.8244
1.9666	37.5	150	1.7812
2.686	50.0	200	1.7117
1.8873	62.5	250	1.6662
2.0797	75.0	300	1.6360
2.2612	87.5	350	1.6419
1.954	100.0	400	1.6110
2.0358	112.5	450	1.6159
1.9712	125.0	500	1.6164
2.1658	137.5	550	1.6242
2.7702	150.0	600	1.6097
1.9429	162.5	650	1.6083
1.947	175.0	700	1.5989
2.0561	187.5	750	1.6029
2.0323	200.0	800	1.5877
2.0326	212.5	850	1.5870
1.7113	225.0	900	1.5835
1.6647	237.5	950	1.5811
2.2978	250.0	1000	1.5814

Framework versions

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