2024-01-10_one_stage_subgraphs_weighted_txt_vis_conc_1_4_8_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: 1.2492
• Accuracy: 0.7825
• Exit 0 Accuracy: 0.2825
• Exit 1 Accuracy: 0.48
• Exit 2 Accuracy: 0.6675
• Exit 3 Accuracy: 0.7675
• Exit 4 Accuracy: 0.7825

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	2.6704	0.1675	0.065	0.0875	0.0625	0.0625	0.105
No log	1.98	33	2.4843	0.2425	0.1075	0.1425	0.0625	0.0625	0.17
No log	3.0	50	2.3121	0.295	0.1325	0.1375	0.0625	0.0625	0.2575
No log	3.96	66	2.0499	0.4025	0.145	0.11	0.0625	0.0625	0.31
No log	4.98	83	1.7740	0.5425	0.14	0.1225	0.0625	0.0625	0.4175
No log	6.0	100	1.4803	0.6225	0.1525	0.1125	0.0625	0.0625	0.5075
No log	6.96	116	1.3264	0.6625	0.1625	0.1375	0.0675	0.0625	0.5775
No log	7.98	133	1.1949	0.705	0.1725	0.13	0.0775	0.0625	0.6225
No log	9.0	150	1.0490	0.73	0.1675	0.15	0.1175	0.0625	0.6875
No log	9.96	166	0.9819	0.7375	0.185	0.1825	0.1225	0.0625	0.685
No log	10.98	183	0.9539	0.74	0.1825	0.18	0.155	0.0625	0.695
No log	12.0	200	0.8850	0.7675	0.195	0.2275	0.1925	0.07	0.7475
No log	12.96	216	0.8869	0.75	0.1925	0.225	0.3	0.1125	0.75
No log	13.98	233	0.9250	0.7475	0.2025	0.255	0.325	0.12	0.75
No log	15.0	250	0.8685	0.7875	0.215	0.18	0.315	0.14	0.78
No log	15.96	266	0.8504	0.7875	0.2375	0.2225	0.405	0.405	0.79
No log	16.98	283	0.9215	0.7725	0.235	0.1975	0.355	0.5075	0.775
No log	18.0	300	0.9816	0.7575	0.2575	0.2325	0.405	0.5825	0.7575
No log	18.96	316	0.9900	0.755	0.255	0.2075	0.4525	0.5925	0.7675
No log	19.98	333	0.9651	0.78	0.255	0.245	0.485	0.595	0.7825
No log	21.0	350	1.0235	0.7525	0.24	0.3	0.465	0.6825	0.7625
No log	21.96	366	1.0137	0.7875	0.24	0.345	0.4775	0.7275	0.785
No log	22.98	383	1.0876	0.765	0.235	0.345	0.4675	0.74	0.765
No log	24.0	400	1.0696	0.77	0.2525	0.3025	0.52	0.755	0.775
No log	24.96	416	1.0440	0.775	0.2525	0.285	0.49	0.77	0.7725
No log	25.98	433	1.0962	0.76	0.255	0.2825	0.4975	0.775	0.7625
No log	27.0	450	1.1214	0.7725	0.275	0.3525	0.515	0.7775	0.7725
No log	27.96	466	1.1593	0.7775	0.27	0.325	0.52	0.7625	0.7775
No log	28.98	483	1.1341	0.7625	0.2725	0.3925	0.545	0.7625	0.7625
0.5624	30.0	500	1.1682	0.7675	0.2775	0.415	0.5875	0.76	0.77
0.5624	30.96	516	1.1978	0.7725	0.26	0.415	0.585	0.7675	0.77
0.5624	31.98	533	1.1051	0.78	0.26	0.4275	0.59	0.7775	0.78
0.5624	33.0	550	1.0934	0.78	0.25	0.41	0.5825	0.775	0.7825
0.5624	33.96	566	1.1564	0.7825	0.26	0.395	0.5925	0.7775	0.78
0.5624	34.98	583	1.1605	0.7825	0.2775	0.4425	0.615	0.77	0.785
0.5624	36.0	600	1.1793	0.775	0.2825	0.4325	0.6	0.7775	0.775
0.5624	36.96	616	1.1635	0.785	0.29	0.4375	0.61	0.7625	0.7825
0.5624	37.98	633	1.1591	0.775	0.28	0.4375	0.615	0.765	0.7775
0.5624	39.0	650	1.1568	0.7775	0.2875	0.455	0.625	0.765	0.7775
0.5624	39.96	666	1.1686	0.78	0.2825	0.4375	0.6275	0.7675	0.78
0.5624	40.98	683	1.1720	0.785	0.275	0.45	0.6275	0.77	0.785
0.5624	42.0	700	1.1977	0.785	0.2775	0.4425	0.6375	0.76	0.785
0.5624	42.96	716	1.2252	0.7825	0.275	0.4575	0.6325	0.755	0.785
0.5624	43.98	733	1.2122	0.79	0.28	0.4625	0.64	0.76	0.79
0.5624	45.0	750	1.2193	0.78	0.2875	0.4625	0.6525	0.7675	0.775
0.5624	45.96	766	1.2197	0.7825	0.285	0.46	0.66	0.755	0.7775
0.5624	46.98	783	1.1791	0.785	0.2825	0.47	0.6475	0.7625	0.785
0.5624	48.0	800	1.1879	0.79	0.2825	0.47	0.655	0.7625	0.7925
0.5624	48.96	816	1.1847	0.795	0.285	0.4725	0.6525	0.7625	0.7975
0.5624	49.98	833	1.1964	0.7925	0.2825	0.48	0.665	0.765	0.785
0.5624	51.0	850	1.2254	0.7825	0.285	0.47	0.665	0.7675	0.7825
0.5624	51.96	866	1.2455	0.7875	0.285	0.4775	0.665	0.7625	0.785
0.5624	52.98	883	1.2492	0.7875	0.2825	0.48	0.6675	0.7625	0.7875
0.5624	54.0	900	1.2459	0.785	0.285	0.47	0.67	0.77	0.785
0.5624	54.96	916	1.2453	0.7825	0.2825	0.475	0.665	0.7675	0.7825
0.5624	55.98	933	1.2505	0.785	0.28	0.4775	0.665	0.7625	0.785
0.5624	57.0	950	1.2494	0.7825	0.2825	0.48	0.6675	0.765	0.7825
0.5624	57.6	960	1.2492	0.7825	0.2825	0.48	0.6675	0.7675	0.7825

Framework versions

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