Training Loss	Epoch	Step	Validation Loss	Accuracy	Brier Loss	Nll	F1 Micro	F1 Macro	Ece	Aurc
No log	1.0	250	5.3199	0.2082	0.8704	3.2359	0.2082	0.1578	0.0869	0.5991
6.235	2.0	500	4.4722	0.336	0.7969	6.1930	0.336	0.3239	0.0602	0.4723
6.235	3.0	750	4.4441	0.3252	0.8299	7.3074	0.3252	0.2997	0.1510	0.4987
3.7524	4.0	1000	5.8119	0.2512	0.9906	9.0519	0.2512	0.2274	0.3051	0.6464
3.7524	5.0	1250	3.3062	0.4848	0.6850	5.3438	0.4848	0.4758	0.1312	0.3066
2.6846	6.0	1500	2.5489	0.5737	0.5723	4.2350	0.5737	0.5633	0.0986	0.2059
2.6846	7.0	1750	1.9242	0.646	0.4801	3.5335	0.646	0.6475	0.0736	0.1432
1.7535	8.0	2000	1.5612	0.681	0.4321	3.0209	0.681	0.6849	0.0656	0.1183
1.7535	9.0	2250	1.6465	0.6943	0.4279	3.0835	0.6943	0.6880	0.0740	0.1125
1.2416	10.0	2500	1.4493	0.6963	0.4166	2.7813	0.6963	0.7028	0.0650	0.1096
1.2416	11.0	2750	1.3074	0.724	0.3870	2.6976	0.724	0.7209	0.0586	0.0978
0.9163	12.0	3000	1.4593	0.7175	0.4025	2.7043	0.7175	0.7181	0.0565	0.1089
0.9163	13.0	3250	1.3412	0.7105	0.4017	2.6889	0.7105	0.7090	0.0701	0.1062
0.7085	14.0	3500	1.1755	0.7342	0.3648	2.4382	0.7342	0.7368	0.0612	0.0857
0.7085	15.0	3750	1.2057	0.7298	0.3733	2.4317	0.7298	0.7346	0.0591	0.0898
0.563	16.0	4000	1.2093	0.7288	0.3730	2.4142	0.7288	0.7329	0.0485	0.0914
0.563	17.0	4250	1.2197	0.7392	0.3636	2.5076	0.7392	0.7381	0.0503	0.0852
0.4589	18.0	4500	1.3051	0.74	0.3646	2.5211	0.74	0.7361	0.0455	0.0860
0.4589	19.0	4750	1.3057	0.7415	0.3636	2.4452	0.7415	0.7389	0.0459	0.0858
0.3787	20.0	5000	1.2422	0.7435	0.3577	2.5107	0.7435	0.7417	0.0474	0.0835
0.3787	21.0	5250	1.3069	0.7408	0.3645	2.5522	0.7408	0.7384	0.0450	0.0875
0.3203	22.0	5500	1.0779	0.7555	0.3449	2.3048	0.7555	0.7556	0.0514	0.0786
0.3203	23.0	5750	1.1117	0.7522	0.3481	2.3305	0.7522	0.7559	0.0501	0.0790
0.2752	24.0	6000	1.1987	0.7545	0.3504	2.4084	0.7545	0.7521	0.0468	0.0803
0.2752	25.0	6250	1.0761	0.7528	0.3434	2.3344	0.7528	0.7548	0.0489	0.0772
0.2413	26.0	6500	1.0694	0.7585	0.3385	2.3028	0.7585	0.7612	0.0521	0.0750
0.2413	27.0	6750	1.0564	0.758	0.3401	2.2827	0.7580	0.7601	0.0450	0.0760
0.2108	28.0	7000	1.0964	0.756	0.3424	2.2600	0.756	0.7583	0.0442	0.0775
0.2108	29.0	7250	1.0759	0.7515	0.3376	2.2334	0.7515	0.7555	0.0477	0.0753
0.1891	30.0	7500	1.0712	0.758	0.3399	2.3159	0.7580	0.7614	0.0432	0.0765
0.1891	31.0	7750	1.0485	0.7552	0.3349	2.2442	0.7552	0.7536	0.0461	0.0738
0.1697	32.0	8000	1.0812	0.7622	0.3362	2.3025	0.7622	0.7609	0.0468	0.0741
0.1697	33.0	8250	1.0069	0.765	0.3294	2.2406	0.765	0.7649	0.0441	0.0712
0.1549	34.0	8500	1.0517	0.7622	0.3339	2.2801	0.7622	0.7618	0.0394	0.0731
0.1549	35.0	8750	1.0146	0.761	0.3322	2.2421	0.761	0.7611	0.0434	0.0728
0.1403	36.0	9000	1.0122	0.7648	0.3308	2.2229	0.7648	0.7662	0.0460	0.0715
0.1403	37.0	9250	1.0170	0.7615	0.3308	2.2247	0.7615	0.7632	0.0381	0.0721
0.1277	38.0	9500	1.0135	0.763	0.3310	2.2289	0.763	0.7633	0.0412	0.0717
0.1277	39.0	9750	1.0230	0.7595	0.3323	2.2202	0.7595	0.7609	0.0450	0.0727
0.1163	40.0	10000	1.0133	0.7622	0.3319	2.2613	0.7622	0.7637	0.0429	0.0723
0.1163	41.0	10250	1.0142	0.7638	0.3319	2.2730	0.7638	0.7629	0.0430	0.0724
0.1063	42.0	10500	0.9957	0.7618	0.3297	2.2467	0.7618	0.7627	0.0425	0.0717
0.1063	43.0	10750	1.0029	0.7658	0.3280	2.2498	0.7658	0.7672	0.0464	0.0705
0.0974	44.0	11000	0.9954	0.766	0.3287	2.2435	0.766	0.7669	0.0443	0.0709
0.0974	45.0	11250	0.9887	0.768	0.3284	2.2407	0.768	0.7687	0.0340	0.0707
0.0914	46.0	11500	0.9890	0.7668	0.3276	2.2161	0.7668	0.7669	0.0433	0.0704
0.0914	47.0	11750	0.9854	0.766	0.3270	2.2357	0.766	0.7668	0.0360	0.0703
0.0854	48.0	12000	0.9868	0.7675	0.3273	2.2340	0.7675	0.7681	0.0420	0.0705
0.0854	49.0	12250	0.9843	0.765	0.3267	2.2427	0.765	0.7654	0.0405	0.0702
0.0824	50.0	12500	0.9849	0.7652	0.3272	2.2207	0.7652	0.7659	0.0400	0.0704

Framework versions

Transformers 4.26.1
Pytorch 1.13.1.post200
Datasets 2.9.0
Tokenizers 0.13.2

jordyvl/resnet101_rvl-cdip-cnn_rvl_cdip-NK1000_kd_MSE_rand

resnet101_rvl-cdip-cnn_rvl_cdip-NK1000_kd_MSE_rand

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

jordyvl/
resnet101_rvl-cdip-cnn_rvl_cdip-NK1000_kd_MSE_rand