EElayoutlmv3_jordyvl_rvl_cdip_100_examples_per_class_2023-12-04_txt_vis_concat_enc_7_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: 0.9969
• Accuracy: 0.7375
• Exit 0 Accuracy: 0.0575
• Exit 1 Accuracy: 0.135

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	2.7554	0.09	0.04	0.07
No log	1.96	8	2.7058	0.15	0.0475	0.0625
No log	2.96	12	2.6226	0.1675	0.05	0.0625
No log	3.96	16	2.5435	0.19	0.0525	0.0625
No log	4.96	20	2.4855	0.2075	0.0525	0.0625
No log	5.96	24	2.3833	0.2925	0.0525	0.065
No log	6.96	28	2.2843	0.3375	0.0525	0.0725
No log	7.96	32	2.1867	0.3525	0.0525	0.0875
No log	8.96	36	2.0982	0.3775	0.0525	0.09
No log	9.96	40	1.9870	0.4375	0.055	0.09
No log	10.96	44	1.8947	0.4575	0.055	0.0725
No log	11.96	48	1.7869	0.495	0.055	0.065
No log	12.96	52	1.7034	0.54	0.0575	0.0575
No log	13.96	56	1.6100	0.57	0.0575	0.055
No log	14.96	60	1.5372	0.5675	0.0575	0.065
No log	15.96	64	1.4592	0.61	0.0575	0.07
No log	16.96	68	1.4024	0.6125	0.0575	0.08
No log	17.96	72	1.3378	0.64	0.0575	0.075
No log	18.96	76	1.2934	0.66	0.0575	0.085
No log	19.96	80	1.2202	0.655	0.0575	0.09
No log	20.96	84	1.1826	0.6725	0.055	0.085
No log	21.96	88	1.1325	0.6925	0.055	0.09
No log	22.96	92	1.0960	0.7275	0.055	0.0875
No log	23.96	96	1.0852	0.69	0.055	0.0925
No log	24.96	100	1.0503	0.7075	0.055	0.0875
No log	25.96	104	1.0417	0.705	0.055	0.0975
No log	26.96	108	1.0174	0.71	0.055	0.095
No log	27.96	112	0.9902	0.715	0.055	0.1
No log	28.96	116	1.0088	0.6975	0.055	0.1025
No log	29.96	120	0.9992	0.7125	0.055	0.1075
No log	30.96	124	0.9688	0.72	0.0575	0.1125
No log	31.96	128	0.9745	0.7275	0.0575	0.11
No log	32.96	132	0.9688	0.7175	0.055	0.1125
No log	33.96	136	0.9720	0.7225	0.055	0.11
No log	34.96	140	0.9515	0.7375	0.055	0.1125
No log	35.96	144	0.9717	0.73	0.055	0.115
No log	36.96	148	0.9686	0.7225	0.055	0.1125
No log	37.96	152	0.9646	0.7275	0.055	0.115
No log	38.96	156	0.9522	0.7375	0.055	0.1225
No log	39.96	160	0.9844	0.735	0.055	0.12
No log	40.96	164	0.9815	0.745	0.055	0.1075
No log	41.96	168	0.9866	0.7225	0.055	0.115
No log	42.96	172	0.9835	0.7325	0.055	0.115
No log	43.96	176	0.9902	0.74	0.055	0.11
No log	44.96	180	0.9843	0.745	0.055	0.115
No log	45.96	184	1.0099	0.725	0.055	0.1225
No log	46.96	188	0.9917	0.7325	0.0575	0.125
No log	47.96	192	0.9817	0.735	0.0575	0.13
No log	48.96	196	1.0048	0.73	0.055	0.13
No log	49.96	200	0.9973	0.7375	0.0575	0.1275
No log	50.96	204	0.9899	0.7375	0.0575	0.1275
No log	51.96	208	0.9913	0.7375	0.0575	0.1275
No log	52.96	212	0.9975	0.735	0.0575	0.13
No log	53.96	216	1.0028	0.7325	0.0575	0.13
No log	54.96	220	0.9928	0.735	0.0575	0.1275
No log	55.96	224	0.9866	0.735	0.0575	0.1325
No log	56.96	228	0.9884	0.7375	0.0575	0.1275
No log	57.96	232	0.9943	0.7425	0.0575	0.13
No log	58.96	236	0.9967	0.74	0.0575	0.13
No log	59.96	240	0.9969	0.7375	0.0575	0.135

Framework versions

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