resnet101_rvl-cdip-cnn_rvl_cdip-NK1000_hint_rand

This model is a fine-tuned version of bdpc/resnet101_rvl-cdip on the None dataset. It achieves the following results on the evaluation set:

• Loss: 19.9624
• Accuracy: 0.765
• Brier Loss: 0.3910
• Nll: 2.2998
• F1 Micro: 0.765
• F1 Macro: 0.7641
• Ece: 0.1669
• Aurc: 0.0775

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: 0.0001
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Brier Loss	Nll	F1 Micro	F1 Macro	Ece	Aurc
No log	1.0	250	26.8546	0.2102	0.8652	3.4225	0.2102	0.1897	0.0659	0.6137
26.8891	2.0	500	25.9803	0.2845	0.8381	3.4360	0.2845	0.2457	0.0762	0.5573
26.8891	3.0	750	25.8252	0.337	0.8092	3.6483	0.337	0.3168	0.1091	0.4988
25.1656	4.0	1000	24.9957	0.4088	0.7588	3.0528	0.4088	0.3859	0.1283	0.4163
25.1656	5.0	1250	24.2209	0.5517	0.5964	2.6661	0.5517	0.5510	0.0726	0.2489
23.8526	6.0	1500	23.4086	0.5915	0.5431	2.5360	0.5915	0.5840	0.0549	0.1994
23.8526	7.0	1750	23.0800	0.625	0.5049	2.4839	0.625	0.6268	0.0608	0.1668
22.7511	8.0	2000	22.9512	0.6573	0.4660	2.4202	0.6573	0.6564	0.0565	0.1407
22.7511	9.0	2250	22.7991	0.6783	0.4509	2.4137	0.6783	0.6749	0.0634	0.1315
21.9881	10.0	2500	22.8533	0.6352	0.5056	2.6549	0.6352	0.6231	0.1085	0.1554
21.9881	11.0	2750	22.7499	0.669	0.4673	2.5291	0.669	0.6642	0.1053	0.1347
21.391	12.0	3000	22.6520	0.6757	0.4767	2.5038	0.6757	0.6745	0.1204	0.1355
21.391	13.0	3250	22.4767	0.6737	0.4850	2.6030	0.6737	0.6718	0.1385	0.1380
20.9347	14.0	3500	22.3023	0.6767	0.4832	2.5438	0.6767	0.6770	0.1594	0.1301
20.9347	15.0	3750	22.1482	0.693	0.4666	2.5622	0.693	0.6913	0.1581	0.1209
20.5776	16.0	4000	22.1655	0.6943	0.4849	2.5685	0.6943	0.6994	0.1766	0.1288
20.5776	17.0	4250	22.0213	0.686	0.4922	2.6576	0.686	0.6925	0.1749	0.1250
20.2836	18.0	4500	21.5434	0.7023	0.4560	2.5508	0.7023	0.7018	0.1720	0.1146
20.2836	19.0	4750	21.7105	0.715	0.4501	2.5953	0.715	0.7128	0.1738	0.1083
20.0339	20.0	5000	21.6301	0.7057	0.4645	2.6131	0.7057	0.7033	0.1701	0.1153
20.0339	21.0	5250	21.9130	0.7007	0.4825	2.7114	0.7007	0.6989	0.1917	0.1231
19.8026	22.0	5500	21.7975	0.713	0.4702	2.7340	0.713	0.7117	0.1879	0.1148
19.8026	23.0	5750	21.5577	0.7173	0.4621	2.7138	0.7173	0.7100	0.1931	0.1072
19.6001	24.0	6000	21.2486	0.722	0.4491	2.5651	0.722	0.7214	0.1853	0.1045
19.6001	25.0	6250	21.0363	0.7348	0.4344	2.4688	0.7348	0.7364	0.1780	0.0974
19.4158	26.0	6500	21.3527	0.728	0.4495	2.7492	0.728	0.7219	0.1864	0.1005
19.4158	27.0	6750	20.8258	0.7355	0.4339	2.4375	0.7355	0.7352	0.1838	0.0943
19.2585	28.0	7000	21.0491	0.729	0.4465	2.6324	0.729	0.7245	0.1953	0.1010
19.2585	29.0	7250	20.7774	0.7425	0.4283	2.4694	0.7425	0.7410	0.1799	0.0908
19.1051	30.0	7500	20.6908	0.741	0.4311	2.4924	0.7410	0.7405	0.1890	0.0888
19.1051	31.0	7750	20.8242	0.743	0.4264	2.5098	0.743	0.7407	0.1826	0.0903
18.9722	32.0	8000	20.6257	0.7435	0.4288	2.4740	0.7435	0.7432	0.1865	0.0872
18.9722	33.0	8250	20.6265	0.745	0.4289	2.4552	0.745	0.7435	0.1862	0.0929
18.854	34.0	8500	20.4251	0.7505	0.4124	2.4631	0.7505	0.7513	0.1790	0.0845
18.854	35.0	8750	20.4164	0.741	0.4278	2.3888	0.7410	0.7402	0.1859	0.0889
18.7477	36.0	9000	20.3432	0.751	0.4184	2.4020	0.751	0.7485	0.1800	0.0850
18.7477	37.0	9250	20.4310	0.7555	0.4154	2.4639	0.7555	0.7528	0.1759	0.0842
18.6548	38.0	9500	20.1987	0.7542	0.4111	2.2921	0.7542	0.7542	0.1792	0.0815
18.6548	39.0	9750	20.2326	0.7562	0.4017	2.3536	0.7562	0.7537	0.1767	0.0829
18.5776	40.0	10000	20.1571	0.7575	0.3985	2.3405	0.7575	0.7568	0.1703	0.0811
18.5776	41.0	10250	20.1580	0.7625	0.3962	2.3855	0.7625	0.7621	0.1713	0.0814
18.5133	42.0	10500	20.0952	0.7572	0.4038	2.3600	0.7572	0.7563	0.1768	0.0794
18.5133	43.0	10750	20.1483	0.7575	0.4008	2.3713	0.7575	0.7564	0.1755	0.0820
18.4613	44.0	11000	20.0749	0.762	0.3992	2.3372	0.762	0.7618	0.1720	0.0795
18.4613	45.0	11250	20.0664	0.7578	0.4035	2.3570	0.7577	0.7566	0.1769	0.0795
18.4218	46.0	11500	19.9611	0.7622	0.3946	2.3399	0.7622	0.7617	0.1674	0.0784
18.4218	47.0	11750	19.9678	0.7632	0.3907	2.3011	0.7632	0.7635	0.1692	0.0772
18.3945	48.0	12000	19.9950	0.763	0.3910	2.2773	0.763	0.7616	0.1695	0.0775
18.3945	49.0	12250	20.0013	0.7625	0.3911	2.2875	0.7625	0.7618	0.1705	0.0777
18.3792	50.0	12500	19.9624	0.765	0.3910	2.2998	0.765	0.7641	0.1669	0.0775

Framework versions

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