layoutlmv3-finetuned

This model is a fine-tuned version of microsoft/layoutlmv3-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.3160
• Precision: 0.8865
• Recall: 0.8880
• F1: 0.8872
• Accuracy: 0.9313

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: 4000

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	0.3155	100	0.3799	0.8272	0.8281	0.8276	0.8989
No log	0.6309	200	0.3640	0.8102	0.8403	0.825	0.8976
No log	0.9464	300	0.3647	0.8169	0.8445	0.8305	0.8983
No log	1.2618	400	0.3327	0.8465	0.8486	0.8475	0.9095
0.3176	1.5773	500	0.3260	0.8389	0.8509	0.8449	0.9106
0.3176	1.8927	600	0.3263	0.8450	0.8489	0.8469	0.9090
0.3176	2.2082	700	0.3157	0.8531	0.8593	0.8562	0.9165
0.3176	2.5237	800	0.3015	0.8532	0.8665	0.8598	0.9157
0.3176	2.8391	900	0.2953	0.8486	0.8644	0.8565	0.9170
0.2295	3.1546	1000	0.3092	0.8611	0.8731	0.8671	0.9191
0.2295	3.4700	1100	0.2928	0.8707	0.8695	0.8701	0.9227
0.2295	3.7855	1200	0.2885	0.8650	0.8725	0.8688	0.9230
0.2295	4.1009	1300	0.2968	0.8727	0.8717	0.8722	0.9221
0.2295	4.4164	1400	0.3031	0.8654	0.8714	0.8684	0.9201
0.1703	4.7319	1500	0.3076	0.8628	0.8725	0.8676	0.9188
0.1703	5.0473	1600	0.2884	0.8791	0.8710	0.8751	0.9251
0.1703	5.3628	1700	0.3150	0.8669	0.8763	0.8716	0.9216
0.1703	5.6782	1800	0.3061	0.8634	0.8812	0.8722	0.9232
0.1703	5.9937	1900	0.2930	0.8776	0.8785	0.8780	0.9264
0.1333	6.3091	2000	0.3095	0.8726	0.8804	0.8765	0.9255
0.1333	6.6246	2100	0.2997	0.8757	0.8801	0.8779	0.9262
0.1333	6.9401	2200	0.3002	0.8783	0.8801	0.8792	0.9278
0.1333	7.2555	2300	0.2980	0.8795	0.8837	0.8816	0.9289
0.1333	7.5710	2400	0.3057	0.8813	0.8822	0.8818	0.9282
0.1112	7.8864	2500	0.3050	0.8799	0.8791	0.8795	0.9280
0.1112	8.2019	2600	0.3030	0.8819	0.8819	0.8819	0.9296
0.1112	8.5174	2700	0.3190	0.8664	0.8831	0.8747	0.9249
0.1112	8.8328	2800	0.3137	0.8821	0.8822	0.8821	0.9283
0.1112	9.1483	2900	0.3110	0.8883	0.8820	0.8851	0.9308
0.0895	9.4637	3000	0.3184	0.8769	0.8851	0.8809	0.9283
0.0895	9.7792	3100	0.3067	0.8769	0.8891	0.8829	0.9294
0.0895	10.0946	3200	0.3161	0.8819	0.8871	0.8845	0.9306
0.0895	10.4101	3300	0.3251	0.8762	0.8874	0.8818	0.9280
0.0895	10.7256	3400	0.3123	0.8863	0.8851	0.8857	0.9309
0.0788	11.0410	3500	0.3160	0.8865	0.8880	0.8872	0.9313
0.0788	11.3565	3600	0.3205	0.8835	0.8870	0.8852	0.9303
0.0788	11.6719	3700	0.3249	0.8798	0.8900	0.8849	0.9297
0.0788	11.9874	3800	0.3192	0.8833	0.8874	0.8853	0.9300
0.0788	12.3028	3900	0.3192	0.8838	0.8889	0.8864	0.9303
0.069	12.6183	4000	0.3201	0.8824	0.8895	0.8859	0.9299

Framework versions

• Transformers 4.49.0.dev0
• Pytorch 2.5.1+cu124
• Datasets 3.3.0
• Tokenizers 0.21.0