EElayoutlmv3_jordyvl_rvl_cdip_100_examples_per_class_2023-09-05_txt_vis_con_enc_4_6_7_11_12_ramp

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: 1.7743
• Accuracy: 0.7375
• Exit 0 Accuracy: 0.045
• Exit 1 Accuracy: 0.6925
• Exit 2 Accuracy: 0.735
• Exit 3 Accuracy: 0.74
• Exit 4 Accuracy: 0.74
• Exit 5 Accuracy: 0.7325

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	Exit 5 Accuracy
No log	0.96	16	2.6769	0.175	0.0875	0.0975	0.07	0.0625	0.0625	0.04
No log	1.98	33	2.4757	0.2775	0.0875	0.0825	0.115	0.1	0.0625	0.1275
No log	3.0	50	2.2161	0.36	0.09	0.07	0.1575	0.1025	0.0625	0.22
No log	3.96	66	1.9839	0.4125	0.0825	0.085	0.19	0.1425	0.0625	0.26
No log	4.98	83	1.7415	0.5525	0.085	0.065	0.22	0.135	0.0625	0.34
No log	6.0	100	1.5455	0.61	0.0875	0.065	0.2525	0.2275	0.0625	0.45
No log	6.96	116	1.3759	0.6525	0.085	0.065	0.265	0.345	0.0625	0.4825
No log	7.98	133	1.2670	0.6775	0.0875	0.065	0.33	0.4225	0.0625	0.585
No log	9.0	150	1.1322	0.705	0.085	0.0775	0.36	0.46	0.0625	0.6225
No log	9.96	166	1.0449	0.7325	0.0875	0.09	0.385	0.4775	0.0625	0.655
No log	10.98	183	0.9958	0.7225	0.085	0.085	0.43	0.54	0.0625	0.69
No log	12.0	200	1.0019	0.7075	0.0775	0.085	0.4325	0.56	0.0625	0.7
No log	12.96	216	0.9836	0.7175	0.0725	0.0875	0.4225	0.5925	0.0625	0.71
No log	13.98	233	0.9949	0.7025	0.07	0.095	0.4925	0.55	0.0625	0.6925
No log	15.0	250	0.9548	0.7325	0.085	0.0975	0.5025	0.5925	0.095	0.73
No log	15.96	266	0.9608	0.7475	0.0725	0.105	0.5025	0.6325	0.17	0.7575
No log	16.98	283	0.9872	0.76	0.065	0.11	0.5225	0.6525	0.2975	0.75
No log	18.0	300	1.0311	0.7475	0.065	0.12	0.555	0.67	0.405	0.7575
No log	18.96	316	1.0094	0.7575	0.06	0.155	0.605	0.6725	0.575	0.755
No log	19.98	333	1.0767	0.76	0.0625	0.17	0.63	0.6725	0.64	0.7625
No log	21.0	350	1.1270	0.75	0.065	0.2	0.6525	0.67	0.685	0.745
No log	21.96	366	1.1407	0.745	0.06	0.2175	0.6475	0.6675	0.735	0.7425
No log	22.98	383	1.1239	0.76	0.0725	0.235	0.665	0.69	0.7775	0.765
No log	24.0	400	1.1732	0.7425	0.0675	0.2525	0.685	0.6975	0.745	0.7425
No log	24.96	416	1.2150	0.7575	0.0625	0.265	0.6825	0.7075	0.76	0.755
No log	25.98	433	1.2254	0.765	0.0625	0.27	0.7075	0.7025	0.7675	0.765
No log	27.0	450	1.2767	0.755	0.055	0.31	0.71	0.74	0.7525	0.76
No log	27.96	466	1.2901	0.77	0.055	0.3375	0.7225	0.7425	0.7725	0.7725
No log	28.98	483	1.3303	0.765	0.055	0.4	0.735	0.745	0.7725	0.7675
1.4172	30.0	500	1.4133	0.74	0.0575	0.4225	0.705	0.725	0.7425	0.7425
1.4172	30.96	516	1.4325	0.765	0.06	0.4425	0.725	0.7375	0.7625	0.765
1.4172	31.98	533	1.4553	0.755	0.0575	0.5	0.7125	0.7325	0.7575	0.7525
1.4172	33.0	550	1.4908	0.74	0.06	0.53	0.715	0.7275	0.7375	0.74
1.4172	33.96	566	1.4996	0.7475	0.0575	0.56	0.715	0.735	0.75	0.7475
1.4172	34.98	583	1.5083	0.75	0.06	0.5825	0.735	0.7425	0.755	0.75
1.4172	36.0	600	1.6148	0.74	0.0525	0.6025	0.72	0.735	0.7425	0.74
1.4172	36.96	616	1.5791	0.7525	0.055	0.605	0.74	0.745	0.76	0.7525
1.4172	37.98	633	1.6097	0.745	0.0525	0.6225	0.7325	0.7425	0.745	0.7425
1.4172	39.0	650	1.6481	0.7425	0.055	0.6425	0.73	0.735	0.745	0.745
1.4172	39.96	666	1.6633	0.7475	0.05	0.6625	0.71	0.7325	0.7425	0.7475
1.4172	40.98	683	1.6485	0.7475	0.05	0.6675	0.73	0.7475	0.75	0.75
1.4172	42.0	700	1.7000	0.7425	0.045	0.665	0.73	0.735	0.74	0.74
1.4172	42.96	716	1.7002	0.745	0.045	0.6725	0.725	0.7325	0.745	0.74
1.4172	43.98	733	1.6880	0.7425	0.045	0.6775	0.7325	0.745	0.7425	0.7425
1.4172	45.0	750	1.7557	0.7375	0.0425	0.675	0.7275	0.74	0.7425	0.735
1.4172	45.96	766	1.7474	0.74	0.04	0.68	0.7275	0.74	0.74	0.7375
1.4172	46.98	783	1.7391	0.735	0.0425	0.6875	0.735	0.735	0.7375	0.735
1.4172	48.0	800	1.7523	0.7325	0.0425	0.6925	0.735	0.7375	0.735	0.73
1.4172	48.96	816	1.7304	0.7375	0.0425	0.6875	0.7375	0.735	0.7425	0.7325
1.4172	49.98	833	1.7392	0.74	0.0425	0.69	0.735	0.7425	0.7475	0.7375
1.4172	51.0	850	1.7644	0.74	0.0425	0.6925	0.7375	0.745	0.7425	0.7375
1.4172	51.96	866	1.7633	0.735	0.0425	0.6925	0.735	0.7425	0.7375	0.735
1.4172	52.98	883	1.7486	0.74	0.045	0.6875	0.74	0.745	0.7375	0.735
1.4172	54.0	900	1.7562	0.7325	0.045	0.69	0.7375	0.7425	0.7375	0.7325
1.4172	54.96	916	1.7660	0.735	0.045	0.6925	0.735	0.745	0.7425	0.7325
1.4172	55.98	933	1.7664	0.735	0.045	0.6925	0.735	0.74	0.7425	0.7325
1.4172	57.0	950	1.7739	0.7375	0.045	0.695	0.735	0.74	0.7425	0.7325
1.4172	57.6	960	1.7743	0.7375	0.045	0.6925	0.735	0.74	0.74	0.7325

Framework versions

• Transformers 4.31.0
• Pytorch 2.0.1+cu117
• Datasets 2.13.1
• Tokenizers 0.13.3