EElayoutlmv3_jordyvl_rvl_cdip_100_examples_per_class_2023-12-04_txt_vis_concat_enc_6_gate

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: 0.9813
• Accuracy: 0.7425
• Exit 0 Accuracy: 0.055
• Exit 1 Accuracy: 0.1075

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: 2e-05
• train_batch_size: 8
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 24
• total_train_batch_size: 192
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 60

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Exit 0 Accuracy	Exit 1 Accuracy
No log	0.96	4	2.7555	0.09	0.0425	0.0675
No log	1.96	8	2.7103	0.145	0.0425	0.06
No log	2.96	12	2.6243	0.175	0.0475	0.0625
No log	3.96	16	2.5457	0.19	0.0475	0.06
No log	4.96	20	2.4802	0.22	0.055	0.0725
No log	5.96	24	2.3688	0.29	0.0525	0.1125
No log	6.96	28	2.2693	0.33	0.0525	0.105
No log	7.96	32	2.1806	0.35	0.0525	0.1275
No log	8.96	36	2.0652	0.395	0.0525	0.125
No log	9.96	40	1.9632	0.4325	0.0525	0.1225
No log	10.96	44	1.8904	0.4725	0.05	0.1175
No log	11.96	48	1.7364	0.5575	0.05	0.1225
No log	12.96	52	1.6698	0.5525	0.05	0.12
No log	13.96	56	1.5785	0.59	0.0525	0.115
No log	14.96	60	1.5102	0.5975	0.0575	0.12
No log	15.96	64	1.3949	0.64	0.0575	0.12
No log	16.96	68	1.3608	0.625	0.0575	0.125
No log	17.96	72	1.3010	0.64	0.0575	0.125
No log	18.96	76	1.2554	0.665	0.0575	0.125
No log	19.96	80	1.1876	0.675	0.0575	0.1225
No log	20.96	84	1.1572	0.705	0.0575	0.13
No log	21.96	88	1.1204	0.6875	0.0575	0.1275
No log	22.96	92	1.0787	0.71	0.0575	0.1375
No log	23.96	96	1.0833	0.7075	0.0575	0.13
No log	24.96	100	1.0345	0.725	0.0575	0.13
No log	25.96	104	1.0832	0.7	0.0575	0.14
No log	26.96	108	1.0061	0.7275	0.0575	0.1325
No log	27.96	112	1.0357	0.695	0.0575	0.13
No log	28.96	116	0.9696	0.7325	0.0575	0.1275
No log	29.96	120	1.0087	0.6975	0.0575	0.1275
No log	30.96	124	0.9800	0.72	0.0575	0.1175
No log	31.96	128	0.9802	0.715	0.0575	0.125
No log	32.96	132	0.9751	0.7375	0.0575	0.13
No log	33.96	136	0.9578	0.725	0.0575	0.1275
No log	34.96	140	0.9624	0.725	0.0575	0.1225
No log	35.96	144	0.9676	0.72	0.0575	0.13
No log	36.96	148	0.9572	0.73	0.0575	0.1175
No log	37.96	152	1.0086	0.7175	0.0575	0.125
No log	38.96	156	0.9555	0.735	0.055	0.11
No log	39.96	160	0.9469	0.74	0.055	0.115
No log	40.96	164	0.9835	0.7275	0.055	0.115
No log	41.96	168	0.9364	0.745	0.055	0.1075
No log	42.96	172	0.9590	0.74	0.055	0.105
No log	43.96	176	0.9499	0.7425	0.055	0.1025
No log	44.96	180	0.9731	0.7375	0.055	0.1
No log	45.96	184	0.9719	0.725	0.055	0.1025
No log	46.96	188	0.9669	0.7375	0.055	0.105
No log	47.96	192	0.9713	0.7325	0.055	0.11
No log	48.96	196	0.9738	0.7475	0.055	0.1075
No log	49.96	200	0.9662	0.7425	0.055	0.1025
No log	50.96	204	0.9848	0.73	0.055	0.1025
No log	51.96	208	0.9689	0.7475	0.055	0.11
No log	52.96	212	0.9690	0.7475	0.055	0.105
No log	53.96	216	0.9783	0.7475	0.055	0.1025
No log	54.96	220	0.9831	0.7425	0.055	0.1025
No log	55.96	224	0.9778	0.745	0.055	0.1025
No log	56.96	228	0.9740	0.7475	0.055	0.1025
No log	57.96	232	0.9754	0.7475	0.055	0.1075
No log	58.96	236	0.9793	0.7425	0.055	0.1075
No log	59.96	240	0.9813	0.7425	0.055	0.1075

Framework versions

• Transformers 4.26.1
• Pytorch 1.13.1.post200
• Datasets 2.9.0
• Tokenizers 0.13.2