cdip-tiny_rvl_cdip-NK1000_kd_test

This model is a fine-tuned version of WinKawaks/vit-tiny-patch16-224 on the None dataset. It achieves the following results on the evaluation set:

Loss: 0.4588
Accuracy: 0.8227
Brier Loss: 0.2552
Nll: 1.9117
F1 Micro: 0.8227
F1 Macro: 0.8239
Ece: 0.0507
Aurc: 0.0420

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: 128
eval_batch_size: 128
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	125	1.3371	0.5423	0.5838	2.6217	0.5423	0.5341	0.0583	0.2219
No log	2.0	250	0.9983	0.646	0.4725	2.3143	0.646	0.6407	0.0490	0.1432
No log	3.0	375	0.8094	0.7085	0.3977	2.2571	0.7085	0.7034	0.0500	0.1017
1.2477	4.0	500	0.7633	0.7215	0.3806	2.2013	0.7215	0.7275	0.0447	0.0926
1.2477	5.0	625	0.7295	0.7505	0.3565	2.1741	0.7505	0.7417	0.0631	0.0775
1.2477	6.0	750	0.6706	0.7692	0.3321	2.1869	0.7692	0.7669	0.0614	0.0690
1.2477	7.0	875	0.6933	0.767	0.3344	2.1685	0.767	0.7650	0.0811	0.0676
0.3434	8.0	1000	0.6640	0.7778	0.3251	2.1666	0.7778	0.7795	0.0681	0.0650
0.3434	9.0	1125	0.6874	0.774	0.3328	2.1401	0.774	0.7739	0.0863	0.0660
0.3434	10.0	1250	0.6639	0.7795	0.3194	2.1335	0.7795	0.7800	0.0800	0.0618
0.3434	11.0	1375	0.6827	0.7728	0.3332	2.1140	0.7728	0.7771	0.0891	0.0650
0.1507	12.0	1500	0.6197	0.786	0.3106	2.1052	0.786	0.7873	0.0716	0.0586
0.1507	13.0	1625	0.6264	0.7823	0.3133	2.1077	0.7823	0.7834	0.0784	0.0595
0.1507	14.0	1750	0.5822	0.796	0.2964	2.0567	0.796	0.7983	0.0676	0.0549
0.1507	15.0	1875	0.5900	0.7923	0.3016	2.0704	0.7923	0.7936	0.0724	0.0541
0.107	16.0	2000	0.6044	0.7855	0.3099	2.0625	0.7855	0.7901	0.0730	0.0617
0.107	17.0	2125	0.5692	0.7973	0.2930	2.0627	0.7973	0.7990	0.0676	0.0528
0.107	18.0	2250	0.5836	0.7907	0.2984	2.0575	0.7907	0.7922	0.0749	0.0554
0.107	19.0	2375	0.5469	0.806	0.2835	2.0754	0.806	0.8060	0.0576	0.0498
0.0879	20.0	2500	0.5427	0.804	0.2892	2.0655	0.804	0.8089	0.0593	0.0528
0.0879	21.0	2625	0.5305	0.806	0.2777	2.0213	0.806	0.8070	0.0604	0.0495
0.0879	22.0	2750	0.5146	0.8113	0.2741	2.0127	0.8113	0.8121	0.0534	0.0480
0.0879	23.0	2875	0.5196	0.8107	0.2750	2.0261	0.8108	0.8117	0.0541	0.0489
0.0755	24.0	3000	0.5169	0.8123	0.2743	1.9561	0.8123	0.8127	0.0595	0.0478
0.0755	25.0	3125	0.5129	0.8073	0.2777	2.0020	0.8073	0.8089	0.0552	0.0491
0.0755	26.0	3250	0.4898	0.8177	0.2649	1.9710	0.8178	0.8177	0.0474	0.0451
0.0755	27.0	3375	0.4966	0.8155	0.2682	2.0075	0.8155	0.8163	0.0514	0.0458
0.0652	28.0	3500	0.4883	0.813	0.2690	1.9655	0.813	0.8141	0.0557	0.0465
0.0652	29.0	3625	0.4860	0.8185	0.2659	1.9593	0.8185	0.8194	0.0481	0.0456
0.0652	30.0	3750	0.4760	0.818	0.2600	1.9517	0.818	0.8194	0.0505	0.0441
0.0652	31.0	3875	0.4755	0.8195	0.2611	1.9593	0.8195	0.8196	0.0507	0.0440
0.0568	32.0	4000	0.4763	0.8155	0.2628	1.9508	0.8155	0.8161	0.0484	0.0451
0.0568	33.0	4125	0.4675	0.8225	0.2574	1.9474	0.8225	0.8238	0.0477	0.0433
0.0568	34.0	4250	0.4664	0.8207	0.2579	1.9478	0.8207	0.8220	0.0498	0.0431
0.0568	35.0	4375	0.4635	0.8213	0.2567	1.9233	0.8213	0.8219	0.0481	0.0427
0.0514	36.0	4500	0.4584	0.8245	0.2551	1.9196	0.8245	0.8260	0.0461	0.0424
0.0514	37.0	4625	0.4627	0.825	0.2557	1.9274	0.825	0.8256	0.0454	0.0424
0.0514	38.0	4750	0.4603	0.8213	0.2552	1.9319	0.8213	0.8221	0.0478	0.0425
0.0514	39.0	4875	0.4610	0.8245	0.2560	1.9337	0.8245	0.8252	0.0476	0.0424
0.0483	40.0	5000	0.4603	0.825	0.2559	1.9319	0.825	0.8262	0.0460	0.0421
0.0483	41.0	5125	0.4589	0.8253	0.2545	1.9317	0.8253	0.8260	0.0459	0.0421
0.0483	42.0	5250	0.4586	0.8245	0.2552	1.9192	0.8245	0.8260	0.0524	0.0420
0.0483	43.0	5375	0.4581	0.825	0.2552	1.9179	0.825	0.8263	0.0477	0.0421
0.0465	44.0	5500	0.4573	0.8245	0.2543	1.9187	0.8245	0.8257	0.0457	0.0417
0.0465	45.0	5625	0.4589	0.8225	0.2554	1.9184	0.8225	0.8235	0.0549	0.0421
0.0465	46.0	5750	0.4582	0.823	0.2547	1.9128	0.823	0.8242	0.0512	0.0420
0.0465	47.0	5875	0.4587	0.823	0.2551	1.9135	0.823	0.8241	0.0484	0.0420
0.0458	48.0	6000	0.4585	0.8235	0.2550	1.9127	0.8235	0.8246	0.0479	0.0420
0.0458	49.0	6125	0.4589	0.8227	0.2553	1.9117	0.8227	0.8238	0.0490	0.0421
0.0458	50.0	6250	0.4588	0.8227	0.2552	1.9117	0.8227	0.8239	0.0507	0.0420

Framework versions

Transformers 4.26.1
Pytorch 1.13.1.post200
Datasets 2.9.0
Tokenizers 0.13.2

jordyvl/cdip-tiny_rvl_cdip-NK1000_kd_test

cdip-tiny_rvl_cdip-NK1000_kd_test

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions