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

This model is a fine-tuned version of sumitD/table-transformer-structure-recognition-v1.1-all-finetuned on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1419
• Map: 0.9219
• Map 50: 0.966
• Map 75: 0.9496
• Map Small: -1.0
• Map Medium: 0.8782
• Map Large: 0.921
• Mar 1: 0.5537
• Mar 10: 0.9407
• Mar 100: 0.9694
• Mar Small: -1.0
• Mar Medium: 0.9079
• Mar Large: 0.9693
• Map Table: 0.9882
• Mar 100 Table: 0.9964
• Map Table column: 0.9732
• Mar 100 Table column: 0.9892
• Map Table column header: 0.9543
• Mar 100 Table column header: 0.9847
• Map Table projected row header: 0.8673
• Mar 100 Table projected row header: 0.964
• Map Table row: 0.9584
• Mar 100 Table row: 0.9838
• Map Table spanning cell: 0.7903
• Mar 100 Table spanning cell: 0.8983

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-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• num_epochs: 5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Map	Map 50	Map 75	Map Small	Map Medium	Map Large	Mar 1	Mar 10	Mar 100	Mar Small	Mar Medium	Mar Large	Map Table	Mar 100 Table	Map Table column	Mar 100 Table column	Map Table column header	Mar 100 Table column header	Map Table projected row header	Mar 100 Table projected row header	Map Table row	Mar 100 Table row	Map Table spanning cell	Mar 100 Table spanning cell
0.2338	1.0	23715	0.1991	0.8756	0.9505	0.9307	-1.0	0.7912	0.8748	0.5395	0.9175	0.9471	-1.0	0.8518	0.947	0.9844	0.9935	0.9582	0.981	0.9111	0.9647	0.7701	0.9364	0.9147	0.9577	0.7149	0.8496
0.2048	2.0	47430	0.1915	0.8827	0.9567	0.9384	-1.0	0.8103	0.8823	0.54	0.9197	0.9498	-1.0	0.8538	0.9499	0.9855	0.9944	0.9564	0.9819	0.9047	0.9527	0.7905	0.9437	0.9222	0.9651	0.7371	0.8607
0.1841	3.0	71145	0.1605	0.9087	0.9636	0.9467	-1.0	0.8373	0.9077	0.548	0.933	0.9616	-1.0	0.8868	0.9613	0.9836	0.9935	0.9703	0.9888	0.94	0.9771	0.8468	0.9545	0.9466	0.9781	0.765	0.8778
0.1914	4.0	94860	0.1496	0.9181	0.9652	0.9496	-1.0	0.8741	0.917	0.552	0.9387	0.9678	-1.0	0.9024	0.9676	0.9886	0.9968	0.9724	0.9886	0.9508	0.9824	0.8561	0.9628	0.9574	0.9829	0.7836	0.8934
0.1739	5.0	118575	0.1419	0.9219	0.966	0.9496	-1.0	0.8782	0.921	0.5537	0.9407	0.9694	-1.0	0.9079	0.9693	0.9882	0.9964	0.9732	0.9892	0.9543	0.9847	0.8673	0.964	0.9584	0.9838	0.7903	0.8983

Framework versions

• Transformers 4.48.2
• Pytorch 2.6.0+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0