resnet101_rvl-cdip-cnn_rvl_cdip-NK1000_simkd

This model is a fine-tuned version of microsoft/resnet-50 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.4602
• Accuracy: 0.769
• Brier Loss: 0.3252
• Nll: 2.1002
• F1 Micro: 0.769
• F1 Macro: 0.7667
• Ece: 0.0388
• Aurc: 0.0678

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	1.0910	0.059	0.9372	6.6175	0.059	0.0236	0.0366	0.9408
1.0976	2.0	500	1.0013	0.0838	0.9335	4.2665	0.0838	0.0443	0.0391	0.9208
1.0976	3.0	750	0.9171	0.1335	0.9308	2.8791	0.1335	0.0985	0.0770	0.8928
0.9312	4.0	1000	0.8701	0.1822	0.9243	2.7464	0.1822	0.1497	0.1142	0.8582
0.9312	5.0	1250	0.8306	0.274	0.8635	5.8805	0.274	0.2059	0.1347	0.6733
0.8353	6.0	1500	0.7791	0.396	0.7897	5.0905	0.396	0.3620	0.1762	0.4569
0.8353	7.0	1750	0.7452	0.47	0.7200	4.3882	0.47	0.4357	0.1822	0.3485
0.7569	8.0	2000	0.7148	0.5635	0.6470	3.6418	0.5635	0.5444	0.2022	0.2564
0.7569	9.0	2250	0.6847	0.6092	0.5626	3.0490	0.6092	0.5904	0.1508	0.1932
0.6953	10.0	2500	0.6552	0.648	0.5117	2.7913	0.648	0.6309	0.1312	0.1622
0.6953	11.0	2750	0.6369	0.662	0.4778	2.6400	0.662	0.6468	0.0959	0.1471
0.6357	12.0	3000	0.6074	0.6863	0.4436	2.4974	0.6863	0.6724	0.0734	0.1274
0.6357	13.0	3250	0.5915	0.6975	0.4226	2.4214	0.6975	0.6843	0.0607	0.1173
0.5943	14.0	3500	0.5811	0.7055	0.4080	2.3606	0.7055	0.6923	0.0487	0.1093
0.5943	15.0	3750	0.5694	0.7177	0.3947	2.2689	0.7178	0.7087	0.0553	0.1016
0.5665	16.0	4000	0.5555	0.7225	0.3866	2.2797	0.7225	0.7130	0.0394	0.0981
0.5665	17.0	4250	0.5502	0.725	0.3821	2.2616	0.7250	0.7166	0.0441	0.0957
0.5446	18.0	4500	0.5425	0.7345	0.3704	2.1992	0.7345	0.7277	0.0401	0.0893
0.5446	19.0	4750	0.5325	0.731	0.3670	2.1856	0.731	0.7257	0.0401	0.0872
0.5268	20.0	5000	0.5272	0.738	0.3661	2.2345	0.738	0.7335	0.0467	0.0865
0.5268	21.0	5250	0.5199	0.745	0.3582	2.1676	0.745	0.7407	0.0388	0.0827
0.5107	22.0	5500	0.5146	0.748	0.3530	2.1726	0.748	0.7446	0.0417	0.0802
0.5107	23.0	5750	0.5101	0.7482	0.3516	2.1670	0.7482	0.7445	0.0398	0.0799
0.4973	24.0	6000	0.5076	0.7455	0.3533	2.1814	0.7455	0.7431	0.0396	0.0807
0.4973	25.0	6250	0.4971	0.7512	0.3476	2.1618	0.7513	0.7469	0.0414	0.0780
0.484	26.0	6500	0.4934	0.753	0.3464	2.1725	0.753	0.7497	0.0473	0.0780
0.484	27.0	6750	0.4916	0.756	0.3415	2.1408	0.756	0.7527	0.0480	0.0753
0.4709	28.0	7000	0.4886	0.7582	0.3405	2.1415	0.7582	0.7547	0.0410	0.0746
0.4709	29.0	7250	0.4844	0.7582	0.3377	2.1252	0.7582	0.7556	0.0483	0.0742
0.4617	30.0	7500	0.4831	0.757	0.3372	2.1383	0.757	0.7540	0.0425	0.0731
0.4617	31.0	7750	0.4781	0.759	0.3344	2.1035	0.7590	0.7572	0.0404	0.0718
0.4529	32.0	8000	0.4794	0.7562	0.3375	2.1457	0.7562	0.7545	0.0385	0.0731
0.4529	33.0	8250	0.4777	0.7625	0.3336	2.0834	0.7625	0.7607	0.0433	0.0717
0.4462	34.0	8500	0.4730	0.7598	0.3328	2.1058	0.7598	0.7566	0.0496	0.0716
0.4462	35.0	8750	0.4730	0.761	0.3324	2.0874	0.761	0.7600	0.0461	0.0712
0.4404	36.0	9000	0.4692	0.7635	0.3309	2.0914	0.7635	0.7616	0.0481	0.0703
0.4404	37.0	9250	0.4691	0.7618	0.3298	2.0866	0.7618	0.7598	0.0457	0.0703
0.4351	38.0	9500	0.4666	0.762	0.3294	2.0963	0.762	0.7593	0.0428	0.0700
0.4351	39.0	9750	0.4639	0.7668	0.3265	2.1028	0.7668	0.7652	0.0453	0.0688
0.4309	40.0	10000	0.4627	0.7675	0.3287	2.0981	0.7675	0.7658	0.0449	0.0694
0.4309	41.0	10250	0.4634	0.765	0.3264	2.1151	0.765	0.7631	0.0441	0.0684
0.4269	42.0	10500	0.4626	0.7658	0.3260	2.0977	0.7658	0.7644	0.0414	0.0684
0.4269	43.0	10750	0.4609	0.7672	0.3259	2.0944	0.7672	0.7656	0.0420	0.0681
0.4248	44.0	11000	0.4616	0.7662	0.3253	2.0942	0.7663	0.7652	0.0458	0.0678
0.4248	45.0	11250	0.4605	0.7658	0.3258	2.1447	0.7658	0.7629	0.0408	0.0678
0.4233	46.0	11500	0.4604	0.7662	0.3266	2.1007	0.7663	0.7640	0.0493	0.0686
0.4233	47.0	11750	0.4601	0.7652	0.3252	2.0893	0.7652	0.7633	0.0463	0.0684
0.4221	48.0	12000	0.4600	0.7645	0.3255	2.0695	0.7645	0.7629	0.0472	0.0683
0.4221	49.0	12250	0.4605	0.7662	0.3257	2.0778	0.7663	0.7640	0.0425	0.0682
0.4211	50.0	12500	0.4602	0.769	0.3252	2.1002	0.769	0.7667	0.0388	0.0678

Framework versions

• Transformers 4.33.3
• Pytorch 2.2.0.dev20231002
• Datasets 2.7.1
• Tokenizers 0.13.3