EElayoutlmv3_jordyvl_rvl_cdip_100_examples_per_class_2023-12-04_txt_vis_concat_enc_2_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: 1.0252
• Accuracy: 0.7275
• Exit 0 Accuracy: 0.06
• Exit 1 Accuracy: 0.0625

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	2.7540	0.09	0.04	0.0625
No log	1.96	8	2.7085	0.1475	0.05	0.0625
No log	2.96	12	2.6275	0.16	0.0525	0.0625
No log	3.96	16	2.5536	0.19	0.055	0.0625
No log	4.96	20	2.4924	0.215	0.0525	0.0625
No log	5.96	24	2.3988	0.2675	0.05	0.0625
No log	6.96	28	2.2891	0.3275	0.055	0.07
No log	7.96	32	2.2100	0.3525	0.0575	0.035
No log	8.96	36	2.1345	0.3725	0.055	0.05
No log	9.96	40	2.0877	0.385	0.055	0.0475
No log	10.96	44	1.9486	0.4725	0.055	0.0575
No log	11.96	48	1.8872	0.46	0.055	0.055
No log	12.96	52	1.7546	0.545	0.0575	0.0575
No log	13.96	56	1.6773	0.545	0.0575	0.0625
No log	14.96	60	1.5625	0.5725	0.0575	0.0625
No log	15.96	64	1.4749	0.5975	0.06	0.0625
No log	16.96	68	1.3829	0.6375	0.06	0.0625
No log	17.96	72	1.3381	0.6375	0.06	0.0625
No log	18.96	76	1.2999	0.635	0.06	0.0625
No log	19.96	80	1.2148	0.6875	0.06	0.0625
No log	20.96	84	1.1983	0.6625	0.06	0.0625
No log	21.96	88	1.1637	0.6775	0.06	0.0625
No log	22.96	92	1.1256	0.6875	0.06	0.0625
No log	23.96	96	1.1456	0.6675	0.06	0.0625
No log	24.96	100	1.0709	0.7025	0.06	0.0625
No log	25.96	104	1.1113	0.695	0.06	0.06
No log	26.96	108	1.0416	0.7175	0.06	0.065
No log	27.96	112	1.0716	0.6875	0.06	0.0675
No log	28.96	116	1.0192	0.7175	0.06	0.0625
No log	29.96	120	1.0237	0.705	0.06	0.0625
No log	30.96	124	1.0260	0.7075	0.06	0.0675
No log	31.96	128	0.9777	0.7275	0.06	0.0675
No log	32.96	132	1.0101	0.725	0.06	0.0675
No log	33.96	136	0.9693	0.7225	0.06	0.0725
No log	34.96	140	0.9973	0.725	0.06	0.0725
No log	35.96	144	0.9890	0.7225	0.06	0.0675
No log	36.96	148	0.9947	0.73	0.06	0.0725
No log	37.96	152	1.0048	0.725	0.06	0.0725
No log	38.96	156	0.9622	0.7275	0.06	0.065
No log	39.96	160	0.9894	0.7175	0.06	0.0675
No log	40.96	164	0.9635	0.735	0.06	0.0675
No log	41.96	168	0.9753	0.74	0.06	0.0725
No log	42.96	172	0.9858	0.72	0.06	0.0725
No log	43.96	176	0.9874	0.735	0.06	0.07
No log	44.96	180	0.9856	0.715	0.06	0.065
No log	45.96	184	1.0028	0.7275	0.06	0.0625
No log	46.96	188	1.0067	0.7325	0.06	0.0625
No log	47.96	192	0.9969	0.7275	0.06	0.0625
No log	48.96	196	0.9990	0.74	0.06	0.0625
No log	49.96	200	1.0065	0.735	0.06	0.0625
No log	50.96	204	1.0133	0.735	0.06	0.0625
No log	51.96	208	1.0113	0.735	0.06	0.0625
No log	52.96	212	1.0306	0.7275	0.06	0.0625
No log	53.96	216	1.0203	0.7275	0.06	0.0625
No log	54.96	220	1.0170	0.73	0.06	0.0625
No log	55.96	224	1.0214	0.725	0.06	0.0625
No log	56.96	228	1.0248	0.725	0.06	0.0625
No log	57.96	232	1.0252	0.7275	0.06	0.0625
No log	58.96	236	1.0254	0.7275	0.06	0.0625
No log	59.96	240	1.0252	0.7275	0.06	0.0625

Framework versions

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