EElayoutlmv3_jordyvl_rvl_cdip_100_examples_per_class_2023-07-24_ent_g75

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: 1.8247
• Accuracy: 0.7225
• Exit 0 Accuracy: 0.0725
• Exit 1 Accuracy: 0.0625
• Exit 2 Accuracy: 0.4775
• Exit 3 Accuracy: 0.68
• Exit 4 Accuracy: 0.71

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: 2
• eval_batch_size: 1
• seed: 42
• gradient_accumulation_steps: 24
• total_train_batch_size: 48
• 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	Exit 2 Accuracy	Exit 3 Accuracy	Exit 4 Accuracy
No log	0.96	16	2.6912	0.1525	0.0625	0.0675	0.1075	0.0625	0.0625
No log	1.98	33	2.5217	0.24	0.0725	0.0675	0.1175	0.0675	0.0625
No log	3.0	50	2.3171	0.325	0.085	0.0675	0.1475	0.11	0.085
No log	3.96	66	2.1709	0.3675	0.0625	0.0675	0.2125	0.17	0.12
No log	4.98	83	2.0866	0.4625	0.0625	0.0675	0.2075	0.1825	0.1875
No log	6.0	100	1.8124	0.52	0.065	0.065	0.2175	0.2525	0.3775
No log	6.96	116	1.5919	0.58	0.0675	0.065	0.23	0.3125	0.575
No log	7.98	133	1.3590	0.67	0.0675	0.065	0.2475	0.3425	0.61
No log	9.0	150	1.2656	0.675	0.0675	0.065	0.245	0.405	0.6575
No log	9.96	166	1.1703	0.715	0.07	0.065	0.2575	0.435	0.695
No log	10.98	183	1.0771	0.7075	0.0625	0.065	0.275	0.505	0.685
No log	12.0	200	1.0342	0.73	0.065	0.065	0.285	0.535	0.6925
No log	12.96	216	1.1164	0.695	0.0675	0.065	0.29	0.5475	0.675
No log	13.98	233	1.2252	0.6675	0.075	0.065	0.31	0.5475	0.6875
No log	15.0	250	1.1467	0.7025	0.0625	0.065	0.285	0.5875	0.6875
No log	15.96	266	1.1496	0.705	0.07	0.065	0.3225	0.5775	0.7025
No log	16.98	283	1.2554	0.705	0.065	0.065	0.315	0.59	0.7025
No log	18.0	300	1.1977	0.7075	0.0675	0.065	0.335	0.605	0.705
No log	18.96	316	1.2663	0.71	0.0675	0.065	0.345	0.6125	0.7125
No log	19.98	333	1.2708	0.72	0.0675	0.065	0.3425	0.63	0.72
No log	21.0	350	1.2362	0.7225	0.0675	0.065	0.36	0.63	0.72
No log	21.96	366	1.3372	0.6975	0.0675	0.0625	0.365	0.64	0.715
No log	22.98	383	1.3829	0.7125	0.0725	0.0625	0.3625	0.6375	0.715
No log	24.0	400	1.3804	0.72	0.0725	0.0625	0.3725	0.65	0.715
No log	24.96	416	1.3353	0.725	0.0925	0.0625	0.39	0.6425	0.725
No log	25.98	433	1.4828	0.7125	0.0725	0.0625	0.3925	0.6425	0.7125
No log	27.0	450	1.4568	0.725	0.07	0.0625	0.43	0.655	0.72
No log	27.96	466	1.5439	0.71	0.0675	0.0625	0.415	0.6625	0.705
No log	28.98	483	1.5584	0.7125	0.0725	0.0625	0.44	0.67	0.72
1.8318	30.0	500	1.5625	0.7125	0.0725	0.0625	0.455	0.665	0.7125
1.8318	30.96	516	1.5523	0.7125	0.0675	0.0625	0.445	0.6725	0.71
1.8318	31.98	533	1.6609	0.6925	0.0675	0.0625	0.46	0.675	0.705
1.8318	33.0	550	1.6248	0.72	0.0775	0.0625	0.45	0.675	0.7175
1.8318	33.96	566	1.6584	0.7225	0.0675	0.0625	0.45	0.68	0.7175
1.8318	34.98	583	1.6129	0.7275	0.0675	0.0625	0.4525	0.675	0.725
1.8318	36.0	600	1.6905	0.7225	0.065	0.0625	0.465	0.6775	0.725
1.8318	36.96	616	1.7220	0.7125	0.065	0.0625	0.4525	0.6825	0.71
1.8318	37.98	633	1.6845	0.7175	0.0675	0.0625	0.46	0.685	0.72
1.8318	39.0	650	1.7377	0.7175	0.0725	0.0625	0.4625	0.6775	0.7125
1.8318	39.96	666	1.7170	0.72	0.0875	0.0625	0.4575	0.685	0.7275
1.8318	40.98	683	1.7412	0.7125	0.0725	0.0625	0.465	0.685	0.715
1.8318	42.0	700	1.7694	0.7125	0.07	0.0625	0.4575	0.6825	0.715
1.8318	42.96	716	1.7794	0.715	0.0725	0.0625	0.4675	0.675	0.7125
1.8318	43.98	733	1.7864	0.7175	0.07	0.0625	0.4625	0.6775	0.7175
1.8318	45.0	750	1.8285	0.7075	0.07	0.0625	0.47	0.6775	0.7125
1.8318	45.96	766	1.8063	0.7125	0.07	0.0625	0.4775	0.68	0.71
1.8318	46.98	783	1.8103	0.72	0.0725	0.0625	0.4725	0.6825	0.725
1.8318	48.0	800	1.8323	0.7175	0.0725	0.0625	0.47	0.68	0.72
1.8318	48.96	816	1.8306	0.72	0.0725	0.0625	0.475	0.685	0.715
1.8318	49.98	833	1.8208	0.7225	0.07	0.0625	0.48	0.6825	0.7175
1.8318	51.0	850	1.8062	0.72	0.07	0.0625	0.4775	0.685	0.7175
1.8318	51.96	866	1.8022	0.7175	0.0725	0.0625	0.48	0.68	0.7175
1.8318	52.98	883	1.8090	0.7175	0.07	0.0625	0.475	0.685	0.715
1.8318	54.0	900	1.8197	0.7175	0.0725	0.0625	0.4775	0.68	0.7125
1.8318	54.96	916	1.8260	0.72	0.0725	0.0625	0.4775	0.6825	0.7125
1.8318	55.98	933	1.8257	0.72	0.0725	0.0625	0.4775	0.68	0.7125
1.8318	57.0	950	1.8264	0.72	0.0725	0.0625	0.4775	0.68	0.7125
1.8318	57.6	960	1.8247	0.7225	0.0725	0.0625	0.4775	0.68	0.71

Framework versions

• Transformers 4.31.0
• Pytorch 1.13.1+cu117
• Datasets 2.13.1
• Tokenizers 0.13.3