2024-01-02_one_stage_subgraphs_entropyreg_txt_vis_conc_6_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: 1.1266
• Accuracy: 0.705
• Exit 0 Accuracy: 0.195
• Exit 1 Accuracy: 0.7025

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	2.7544	0.115	0.0575	0.0625
No log	1.96	8	2.6911	0.135	0.125	0.0625
No log	2.96	12	2.6410	0.1775	0.1225	0.18
No log	3.96	16	2.5664	0.2025	0.125	0.1825
No log	4.96	20	2.5036	0.2475	0.1225	0.2475
No log	5.96	24	2.4172	0.28	0.12	0.2275
No log	6.96	28	2.3247	0.3	0.1275	0.2225
No log	7.96	32	2.2355	0.36	0.14	0.2525
No log	8.96	36	2.1384	0.4025	0.1375	0.315
No log	9.96	40	2.0150	0.465	0.14	0.3475
No log	10.96	44	1.9193	0.4925	0.1425	0.37
No log	11.96	48	1.7777	0.5375	0.145	0.4325
No log	12.96	52	1.6960	0.56	0.15	0.5
No log	13.96	56	1.5905	0.59	0.155	0.49
No log	14.96	60	1.5197	0.625	0.155	0.5275
No log	15.96	64	1.4335	0.6475	0.1525	0.5425
No log	16.96	68	1.3831	0.6575	0.1575	0.5675
No log	17.96	72	1.3216	0.6775	0.155	0.575
No log	18.96	76	1.2973	0.6825	0.1575	0.5825
No log	19.96	80	1.2342	0.6975	0.1575	0.6025
No log	20.96	84	1.2190	0.6825	0.16	0.605
No log	21.96	88	1.1758	0.7125	0.1625	0.62
No log	22.96	92	1.1612	0.685	0.1675	0.625
No log	23.96	96	1.1329	0.6925	0.1675	0.64
No log	24.96	100	1.1001	0.7125	0.1675	0.635
No log	25.96	104	1.0943	0.7025	0.175	0.645
No log	26.96	108	1.0794	0.7125	0.18	0.6475
No log	27.96	112	1.0919	0.6925	0.185	0.6475
No log	28.96	116	1.0630	0.72	0.1875	0.6575
No log	29.96	120	1.0831	0.7	0.1875	0.655
No log	30.96	124	1.0581	0.695	0.1875	0.6625
No log	31.96	128	1.0588	0.715	0.1875	0.66
No log	32.96	132	1.0624	0.6975	0.185	0.675
No log	33.96	136	1.0355	0.71	0.1875	0.675
No log	34.96	140	1.0777	0.6925	0.1875	0.665
No log	35.96	144	1.0514	0.71	0.19	0.675
No log	36.96	148	1.0678	0.7	0.1925	0.6825
No log	37.96	152	1.0610	0.7025	0.1925	0.68
No log	38.96	156	1.0726	0.7025	0.195	0.69
No log	39.96	160	1.0818	0.7025	0.195	0.69
No log	40.96	164	1.0893	0.6975	0.1925	0.685
No log	41.96	168	1.0980	0.695	0.195	0.69
No log	42.96	172	1.1009	0.7025	0.1925	0.6925
No log	43.96	176	1.0896	0.705	0.1925	0.695
No log	44.96	180	1.0697	0.7125	0.1925	0.695
No log	45.96	184	1.1185	0.7025	0.1925	0.695
No log	46.96	188	1.0956	0.705	0.1925	0.6925
No log	47.96	192	1.1095	0.71	0.19	0.6975
No log	48.96	196	1.1233	0.7075	0.1925	0.7025
No log	49.96	200	1.1281	0.705	0.1925	0.7025
No log	50.96	204	1.1428	0.6975	0.1925	0.7025
No log	51.96	208	1.1292	0.7025	0.1925	0.71
No log	52.96	212	1.1218	0.7025	0.19	0.7125
No log	53.96	216	1.1143	0.7075	0.1925	0.7025
No log	54.96	220	1.1192	0.7125	0.195	0.7025
No log	55.96	224	1.1338	0.715	0.195	0.7025
No log	56.96	228	1.1333	0.71	0.195	0.7075
No log	57.96	232	1.1291	0.7025	0.195	0.7025
No log	58.96	236	1.1268	0.705	0.195	0.705
No log	59.96	240	1.1266	0.705	0.195	0.7025

Framework versions

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