resnet101_rvl-cdip-_rvl_cdip-NK1000__CEKD_t2.5_a0.5

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: 0.6065
• Accuracy: 0.7915
• Brier Loss: 0.3054
• Nll: 1.9957
• F1 Micro: 0.7915
• F1 Macro: 0.7910
• Ece: 0.0453
• Aurc: 0.0607

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	4.1565	0.1378	0.9318	7.9039	0.1378	0.1073	0.0673	0.8326
4.1485	2.0	500	3.6932	0.3235	0.8832	5.1525	0.3235	0.2725	0.2044	0.5507
4.1485	3.0	750	2.3374	0.4725	0.6611	3.3127	0.4725	0.4311	0.0839	0.2921
2.392	4.0	1000	1.6516	0.588	0.5470	2.8681	0.588	0.5789	0.0620	0.1929
2.392	5.0	1250	1.3260	0.6488	0.4782	2.6378	0.6488	0.6444	0.0486	0.1458
1.1422	6.0	1500	1.0390	0.702	0.4156	2.4086	0.702	0.7029	0.0576	0.1097
1.1422	7.0	1750	0.8420	0.7288	0.3738	2.2222	0.7288	0.7300	0.0553	0.0888
0.708	8.0	2000	0.7753	0.7398	0.3586	2.1518	0.7398	0.7396	0.0587	0.0826
0.708	9.0	2250	0.7797	0.7462	0.3580	2.1095	0.7462	0.7457	0.0581	0.0820
0.5195	10.0	2500	0.7101	0.7602	0.3404	2.0711	0.7602	0.7612	0.0473	0.0733
0.5195	11.0	2750	0.6971	0.7645	0.3338	2.0649	0.7645	0.7653	0.0541	0.0715
0.4176	12.0	3000	0.6936	0.7712	0.3302	2.0265	0.7712	0.7708	0.0515	0.0702
0.4176	13.0	3250	0.6991	0.7662	0.3346	2.0582	0.7663	0.7657	0.0581	0.0723
0.3573	14.0	3500	0.6672	0.7722	0.3246	2.0053	0.7722	0.7723	0.0551	0.0683
0.3573	15.0	3750	0.6735	0.777	0.3244	2.0387	0.777	0.7782	0.0488	0.0671
0.3193	16.0	4000	0.6567	0.776	0.3216	2.0256	0.776	0.7773	0.0499	0.0678
0.3193	17.0	4250	0.6498	0.78	0.3184	1.9865	0.78	0.7802	0.0477	0.0662
0.2893	18.0	4500	0.6763	0.7755	0.3264	2.0844	0.7755	0.7755	0.0531	0.0697
0.2893	19.0	4750	0.6519	0.7815	0.3183	2.0458	0.7815	0.7817	0.0513	0.0658
0.271	20.0	5000	0.6432	0.7823	0.3147	2.0291	0.7823	0.7827	0.0440	0.0645
0.271	21.0	5250	0.6456	0.781	0.3156	2.0493	0.7810	0.7813	0.0487	0.0652
0.2516	22.0	5500	0.6336	0.7823	0.3144	1.9829	0.7823	0.7822	0.0522	0.0642
0.2516	23.0	5750	0.6333	0.7837	0.3128	2.0196	0.7837	0.7836	0.0492	0.0641
0.2397	24.0	6000	0.6337	0.7817	0.3147	2.0180	0.7817	0.7815	0.0494	0.0644
0.2397	25.0	6250	0.6347	0.7857	0.3145	2.0187	0.7857	0.7856	0.0510	0.0641
0.23	26.0	6500	0.6311	0.7815	0.3129	2.0132	0.7815	0.7819	0.0495	0.0637
0.23	27.0	6750	0.6329	0.7853	0.3125	2.0708	0.7853	0.7852	0.0502	0.0635
0.2191	28.0	7000	0.6222	0.786	0.3109	2.0022	0.786	0.7856	0.0483	0.0638
0.2191	29.0	7250	0.6195	0.7863	0.3096	2.0028	0.7863	0.7859	0.0550	0.0620
0.2155	30.0	7500	0.6196	0.7883	0.3090	1.9972	0.7883	0.7883	0.0486	0.0624
0.2155	31.0	7750	0.6167	0.787	0.3080	2.0173	0.787	0.7871	0.0443	0.0623
0.2074	32.0	8000	0.6143	0.7897	0.3073	2.0223	0.7897	0.7893	0.0443	0.0614
0.2074	33.0	8250	0.6123	0.787	0.3078	1.9869	0.787	0.7866	0.0458	0.0619
0.2028	34.0	8500	0.6137	0.7873	0.3070	1.9883	0.7873	0.7868	0.0457	0.0623
0.2028	35.0	8750	0.6152	0.786	0.3085	2.0108	0.786	0.7863	0.0497	0.0626
0.1982	36.0	9000	0.6133	0.7863	0.3077	2.0205	0.7863	0.7862	0.0515	0.0615
0.1982	37.0	9250	0.6145	0.7877	0.3081	1.9930	0.7877	0.7879	0.0444	0.0621
0.1948	38.0	9500	0.6116	0.7857	0.3078	2.0072	0.7857	0.7854	0.0508	0.0619
0.1948	39.0	9750	0.6090	0.788	0.3059	1.9954	0.788	0.7882	0.0430	0.0614
0.1933	40.0	10000	0.6143	0.7897	0.3072	1.9943	0.7897	0.7899	0.0462	0.0618
0.1933	41.0	10250	0.6061	0.7887	0.3041	1.9900	0.7887	0.7889	0.0439	0.0606
0.1882	42.0	10500	0.6070	0.7865	0.3058	1.9907	0.7865	0.7868	0.0438	0.0607
0.1882	43.0	10750	0.6083	0.788	0.3054	2.0095	0.788	0.7877	0.0489	0.0608
0.1871	44.0	11000	0.6083	0.787	0.3054	1.9828	0.787	0.7872	0.0469	0.0607
0.1871	45.0	11250	0.6092	0.7893	0.3057	2.0140	0.7893	0.7891	0.0483	0.0608
0.1862	46.0	11500	0.6057	0.7893	0.3053	2.0064	0.7893	0.7890	0.0450	0.0609
0.1862	47.0	11750	0.6042	0.79	0.3044	1.9691	0.79	0.7899	0.0435	0.0607
0.1845	48.0	12000	0.6068	0.79	0.3053	2.0052	0.79	0.7899	0.0438	0.0608
0.1845	49.0	12250	0.6081	0.7893	0.3062	2.0117	0.7893	0.7890	0.0485	0.0612
0.1836	50.0	12500	0.6065	0.7915	0.3054	1.9957	0.7915	0.7910	0.0453	0.0607

Framework versions

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