2024-01-31_one_stage_subgraphs_weighted_txt_vis_conc_all_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: 1.3452
• Accuracy: 0.7625
• Exit 0 Accuracy: 0.29
• Exit 1 Accuracy: 0.4625
• Exit 2 Accuracy: 0.5225
• Exit 3 Accuracy: 0.585
• Exit 4 Accuracy: 0.625
• Exit 5 Accuracy: 0.695
• Exit 6 Accuracy: 0.71
• Exit 7 Accuracy: 0.73
• Exit 8 Accuracy: 0.73
• Exit 9 Accuracy: 0.7575
• Exit 10 Accuracy: 0.76
• Exit 11 Accuracy: 0.7575
• Exit 12 Accuracy: 0.7625

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	Exit 6 Accuracy	Exit 7 Accuracy	Exit 8 Accuracy	Exit 9 Accuracy	Exit 10 Accuracy	Exit 11 Accuracy	Exit 12 Accuracy
No log	0.96	16	2.6725	0.165	0.0975	0.0625	0.0625	0.0625	0.0625	0.0625	0.0625	0.0625	0.0625	0.0625	0.0625	0.0625	0.0925
No log	1.98	33	2.4536	0.2625	0.125	0.065	0.0625	0.0625	0.0625	0.0625	0.0675	0.0625	0.0625	0.0625	0.0625	0.0625	0.1425
No log	3.0	50	2.1927	0.3825	0.14	0.1025	0.0625	0.0625	0.0625	0.125	0.0975	0.0625	0.0625	0.0625	0.065	0.0625	0.2325
No log	3.96	66	1.9488	0.45	0.1575	0.1025	0.0625	0.0625	0.0625	0.1275	0.13	0.0625	0.0625	0.0625	0.0575	0.0625	0.335
No log	4.98	83	1.6922	0.54	0.16	0.0975	0.0625	0.0625	0.0775	0.24	0.1975	0.12	0.0625	0.0625	0.065	0.0625	0.4175
No log	6.0	100	1.4713	0.615	0.16	0.115	0.0625	0.0625	0.1375	0.3575	0.315	0.2125	0.1075	0.0625	0.075	0.0625	0.4875
No log	6.96	116	1.2943	0.6625	0.165	0.1175	0.0625	0.0675	0.1525	0.3425	0.3675	0.38	0.105	0.0625	0.13	0.095	0.565
No log	7.98	133	1.1118	0.7225	0.175	0.1275	0.0625	0.0675	0.1425	0.36	0.3925	0.4575	0.1075	0.0625	0.16	0.1925	0.6375
No log	9.0	150	1.0188	0.74	0.1725	0.1225	0.0625	0.1025	0.185	0.4075	0.4475	0.535	0.0925	0.0625	0.1025	0.2925	0.6825
No log	9.96	166	0.9689	0.7375	0.17	0.125	0.0625	0.1375	0.165	0.4675	0.475	0.5825	0.0975	0.115	0.2075	0.35	0.6975
No log	10.98	183	0.8788	0.77	0.1775	0.16	0.0625	0.1125	0.2025	0.4975	0.49	0.5975	0.1225	0.1275	0.2525	0.4025	0.75
No log	12.0	200	0.9443	0.7125	0.1725	0.165	0.0625	0.14	0.195	0.515	0.505	0.59	0.1525	0.17	0.2725	0.4	0.705
No log	12.96	216	0.8964	0.7375	0.185	0.2025	0.0675	0.2225	0.21	0.505	0.51	0.6025	0.215	0.1325	0.405	0.4475	0.7375
No log	13.98	233	0.9871	0.73	0.19	0.23	0.065	0.21	0.23	0.4625	0.505	0.6175	0.31	0.14	0.46	0.57	0.72
No log	15.0	250	1.0090	0.73	0.2175	0.265	0.115	0.225	0.2075	0.47	0.525	0.64	0.4625	0.3525	0.485	0.63	0.7375
No log	15.96	266	0.9539	0.76	0.2175	0.2525	0.13	0.185	0.35	0.3975	0.555	0.6375	0.46	0.435	0.46	0.6875	0.7575
No log	16.98	283	0.9204	0.7625	0.2225	0.295	0.1375	0.27	0.3925	0.5075	0.595	0.65	0.5625	0.475	0.5275	0.6825	0.765
No log	18.0	300	0.9639	0.77	0.2375	0.2675	0.17	0.2625	0.455	0.515	0.6175	0.685	0.5775	0.6525	0.63	0.7125	0.765
No log	18.96	316	0.9644	0.7725	0.25	0.3175	0.175	0.3025	0.4975	0.5375	0.65	0.695	0.65	0.66	0.6225	0.745	0.7675
No log	19.98	333	0.9984	0.7675	0.25	0.3275	0.1975	0.295	0.5375	0.58	0.6775	0.7	0.6825	0.65	0.6875	0.7425	0.7625
No log	21.0	350	0.9756	0.775	0.24	0.3025	0.22	0.2825	0.52	0.59	0.6725	0.7125	0.6875	0.6775	0.6975	0.7525	0.7825
No log	21.96	366	1.0060	0.7675	0.235	0.2525	0.255	0.2975	0.545	0.61	0.675	0.7125	0.6925	0.7	0.695	0.75	0.7675
No log	22.98	383	1.0393	0.7675	0.245	0.265	0.2175	0.3125	0.53	0.62	0.69	0.7125	0.7275	0.7125	0.7	0.76	0.7675
No log	24.0	400	1.0382	0.77	0.2475	0.29	0.2475	0.33	0.5525	0.66	0.7	0.72	0.755	0.735	0.7125	0.77	0.7625
No log	24.96	416	1.0630	0.76	0.255	0.2525	0.23	0.3675	0.5325	0.6225	0.685	0.7125	0.75	0.7275	0.7275	0.7675	0.7625
No log	25.98	433	1.0887	0.7625	0.26	0.2825	0.2425	0.3775	0.5325	0.6575	0.7025	0.705	0.7525	0.76	0.755	0.775	0.765
No log	27.0	450	1.1224	0.7675	0.255	0.3125	0.2425	0.3875	0.5275	0.665	0.7	0.7125	0.7475	0.7675	0.75	0.7625	0.7675
No log	27.96	466	1.1230	0.7625	0.275	0.3675	0.2775	0.3825	0.5525	0.67	0.6875	0.7075	0.7425	0.7475	0.745	0.7675	0.7575
No log	28.98	483	1.1384	0.7525	0.2625	0.375	0.3075	0.38	0.5375	0.67	0.7	0.7325	0.745	0.7525	0.7525	0.75	0.75
0.3128	30.0	500	1.1192	0.76	0.285	0.42	0.415	0.4425	0.585	0.6825	0.725	0.7375	0.755	0.7725	0.7625	0.765	0.76
0.3128	30.96	516	1.1687	0.7625	0.27	0.3775	0.335	0.3875	0.5675	0.665	0.685	0.725	0.7375	0.7475	0.7525	0.7575	0.76
0.3128	31.98	533	1.2018	0.755	0.2625	0.37	0.3325	0.385	0.5575	0.6625	0.69	0.73	0.7475	0.76	0.7575	0.76	0.75
0.3128	33.0	550	1.1723	0.7725	0.265	0.355	0.3425	0.4	0.575	0.65	0.685	0.715	0.745	0.76	0.77	0.77	0.775
0.3128	33.96	566	1.2252	0.7475	0.28	0.4175	0.4325	0.4675	0.5775	0.67	0.7025	0.715	0.7475	0.7525	0.76	0.755	0.745
0.3128	34.98	583	1.1831	0.765	0.29	0.4375	0.435	0.4825	0.5925	0.68	0.7175	0.735	0.75	0.7575	0.765	0.765	0.77
0.3128	36.0	600	1.2292	0.755	0.28	0.4375	0.4275	0.4875	0.5875	0.6725	0.7	0.7275	0.745	0.7475	0.745	0.75	0.7525
0.3128	36.96	616	1.2460	0.755	0.2825	0.425	0.435	0.5125	0.59	0.6775	0.7075	0.73	0.745	0.7525	0.755	0.755	0.755
0.3128	37.98	633	1.2560	0.7525	0.2675	0.4525	0.46	0.5175	0.5875	0.6675	0.705	0.7225	0.74	0.745	0.745	0.7475	0.7525
0.3128	39.0	650	1.2463	0.77	0.2825	0.45	0.475	0.5225	0.59	0.67	0.6975	0.73	0.7425	0.76	0.7625	0.76	0.765
0.3128	39.96	666	1.2493	0.765	0.2775	0.455	0.49	0.5325	0.6	0.6825	0.7225	0.7425	0.75	0.765	0.76	0.765	0.7625
0.3128	40.98	683	1.2727	0.7625	0.275	0.47	0.49	0.535	0.61	0.68	0.7	0.7275	0.74	0.7525	0.76	0.76	0.7575
0.3128	42.0	700	1.2951	0.7525	0.2725	0.445	0.495	0.5525	0.5975	0.67	0.6975	0.735	0.7575	0.75	0.7575	0.76	0.75
0.3128	42.96	716	1.2865	0.75	0.275	0.455	0.5075	0.5525	0.6025	0.695	0.71	0.73	0.745	0.7475	0.7575	0.755	0.75
0.3128	43.98	733	1.2864	0.76	0.2775	0.465	0.5025	0.5575	0.6075	0.6925	0.6975	0.73	0.7375	0.7575	0.7625	0.755	0.7575
0.3128	45.0	750	1.3615	0.7575	0.285	0.465	0.5075	0.5525	0.6175	0.6875	0.7075	0.735	0.73	0.7375	0.75	0.745	0.7525
0.3128	45.96	766	1.3161	0.7525	0.2825	0.47	0.5125	0.5575	0.62	0.6825	0.6975	0.7275	0.735	0.7525	0.755	0.755	0.7525
0.3128	46.98	783	1.3508	0.755	0.29	0.4775	0.5125	0.5725	0.6175	0.6875	0.705	0.7175	0.73	0.7525	0.7575	0.755	0.7575
0.3128	48.0	800	1.3321	0.76	0.285	0.47	0.5175	0.565	0.62	0.6925	0.7125	0.73	0.7475	0.755	0.7575	0.7575	0.7575
0.3128	48.96	816	1.3362	0.7625	0.2825	0.465	0.515	0.5725	0.6275	0.69	0.705	0.74	0.745	0.7625	0.765	0.76	0.76
0.3128	49.98	833	1.3070	0.76	0.2825	0.4725	0.5175	0.5725	0.62	0.69	0.71	0.7325	0.7375	0.7525	0.76	0.76	0.7625
0.3128	51.0	850	1.3199	0.7575	0.2875	0.47	0.5125	0.5775	0.625	0.6875	0.705	0.7375	0.735	0.755	0.7675	0.76	0.7575
0.3128	51.96	866	1.3464	0.755	0.2875	0.4675	0.515	0.5775	0.6275	0.685	0.7075	0.73	0.7325	0.755	0.76	0.7525	0.755
0.3128	52.98	883	1.3286	0.7575	0.29	0.47	0.515	0.5775	0.6275	0.6925	0.7125	0.7325	0.7425	0.76	0.7625	0.76	0.76
0.3128	54.0	900	1.3277	0.7625	0.2925	0.4625	0.52	0.5825	0.6275	0.6975	0.715	0.735	0.74	0.7575	0.765	0.7575	0.76
0.3128	54.96	916	1.3274	0.7625	0.2925	0.4625	0.52	0.5875	0.6275	0.695	0.7125	0.7375	0.7375	0.7575	0.7625	0.7625	0.7625
0.3128	55.98	933	1.3393	0.7625	0.29	0.4625	0.5225	0.585	0.625	0.695	0.71	0.7375	0.7375	0.755	0.7625	0.755	0.76
0.3128	57.0	950	1.3453	0.7625	0.29	0.46	0.5225	0.585	0.625	0.695	0.71	0.73	0.73	0.7575	0.76	0.7575	0.7625
0.3128	57.6	960	1.3452	0.7625	0.29	0.4625	0.5225	0.585	0.625	0.695	0.71	0.73	0.73	0.7575	0.76	0.7575	0.7625

Framework versions

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