dit-base-finetuned-rvlcdip-small_rvl_cdip-NK1000_kd_CEKD_t2.5_a0.5

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

Loss: 0.5443
Accuracy: 0.8293
Brier Loss: 0.2695
Nll: 1.2500
F1 Micro: 0.8293
F1 Macro: 0.8301
Ece: 0.0874
Aurc: 0.0693

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: 96
eval_batch_size: 96
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	167	1.1672	0.6082	0.5198	2.4714	0.6082	0.6021	0.0681	0.1731
No log	2.0	334	0.8928	0.7177	0.4028	2.0687	0.7178	0.7161	0.0715	0.0997
1.1617	3.0	501	0.7584	0.7602	0.3454	1.8041	0.7602	0.7637	0.0762	0.0747
1.1617	4.0	668	0.7048	0.768	0.3262	1.6695	0.768	0.7687	0.0550	0.0687
1.1617	5.0	835	0.6921	0.7745	0.3214	1.6428	0.7745	0.7683	0.0477	0.0738
0.4396	6.0	1002	0.6616	0.789	0.3105	1.5655	0.7890	0.7897	0.0586	0.0714
0.4396	7.0	1169	0.6363	0.794	0.3001	1.5694	0.7940	0.7970	0.0549	0.0674
0.4396	8.0	1336	0.6753	0.7792	0.3259	1.4975	0.7792	0.7804	0.0647	0.0777
0.2291	9.0	1503	0.6247	0.8025	0.2979	1.4968	0.8025	0.8037	0.0705	0.0669
0.2291	10.0	1670	0.6347	0.799	0.3032	1.4834	0.799	0.8011	0.0720	0.0743
0.2291	11.0	1837	0.6328	0.7975	0.3045	1.4998	0.7975	0.8031	0.0773	0.0659
0.1575	12.0	2004	0.6442	0.7965	0.3097	1.4447	0.7965	0.7979	0.0714	0.0824
0.1575	13.0	2171	0.6354	0.8013	0.3043	1.4874	0.8013	0.8035	0.0712	0.0741
0.1575	14.0	2338	0.6443	0.799	0.3091	1.5848	0.799	0.8022	0.0791	0.0859
0.1285	15.0	2505	0.6357	0.8017	0.3042	1.5670	0.8017	0.8002	0.0799	0.0685
0.1285	16.0	2672	0.6166	0.807	0.2965	1.4806	0.807	0.8056	0.0720	0.0745
0.1285	17.0	2839	0.6433	0.7993	0.3159	1.5024	0.7993	0.8023	0.0805	0.0857
0.1121	18.0	3006	0.6102	0.8147	0.2960	1.4550	0.8148	0.8144	0.0775	0.0698
0.1121	19.0	3173	0.6616	0.7995	0.3146	1.6009	0.7995	0.7962	0.0892	0.0883
0.1121	20.0	3340	0.6163	0.8037	0.3029	1.4525	0.8037	0.8059	0.0920	0.0771
0.1012	21.0	3507	0.6186	0.8093	0.3017	1.5539	0.8093	0.8111	0.0920	0.0712
0.1012	22.0	3674	0.5982	0.8137	0.2930	1.4533	0.8137	0.8140	0.0815	0.0668
0.1012	23.0	3841	0.5928	0.822	0.2864	1.4312	0.822	0.8218	0.0723	0.0818
0.0888	24.0	4008	0.5931	0.8135	0.2900	1.4129	0.8135	0.8143	0.0894	0.0706
0.0888	25.0	4175	0.5807	0.8183	0.2849	1.4241	0.8183	0.8203	0.0903	0.0683
0.0888	26.0	4342	0.5859	0.8193	0.2869	1.4385	0.8193	0.8194	0.0879	0.0698
0.0828	27.0	4509	0.5957	0.8147	0.2941	1.4132	0.8148	0.8151	0.0847	0.0732
0.0828	28.0	4676	0.5791	0.818	0.2852	1.4231	0.818	0.8185	0.0896	0.0612
0.0828	29.0	4843	0.5888	0.8137	0.2895	1.3998	0.8137	0.8148	0.0925	0.0740
0.0776	30.0	5010	0.5633	0.8225	0.2798	1.3391	0.8225	0.8234	0.0878	0.0760
0.0776	31.0	5177	0.5635	0.8247	0.2785	1.3193	0.8247	0.8256	0.0900	0.0587
0.0776	32.0	5344	0.5580	0.8223	0.2784	1.2970	0.8223	0.8241	0.0905	0.0704
0.0727	33.0	5511	0.5502	0.826	0.2724	1.2733	0.826	0.8268	0.0865	0.0619
0.0727	34.0	5678	0.5448	0.8293	0.2720	1.2237	0.8293	0.8303	0.0820	0.0639
0.0727	35.0	5845	0.5480	0.8257	0.2729	1.2867	0.8257	0.8271	0.0928	0.0586
0.0696	36.0	6012	0.5437	0.8293	0.2703	1.2427	0.8293	0.8298	0.0871	0.0630
0.0696	37.0	6179	0.5460	0.8253	0.2712	1.2629	0.8253	0.8262	0.0912	0.0598
0.0696	38.0	6346	0.5425	0.8295	0.2703	1.2440	0.8295	0.8303	0.0899	0.0611
0.0677	39.0	6513	0.5421	0.8307	0.2690	1.2453	0.8308	0.8319	0.0835	0.0665
0.0677	40.0	6680	0.5406	0.8287	0.2689	1.2465	0.8287	0.8296	0.0895	0.0612
0.0677	41.0	6847	0.5423	0.8277	0.2696	1.2735	0.8277	0.8284	0.0893	0.0604
0.0663	42.0	7014	0.5406	0.8297	0.2676	1.2403	0.8297	0.8306	0.0894	0.0657
0.0663	43.0	7181	0.5410	0.8313	0.2686	1.2359	0.8313	0.8323	0.0895	0.0635
0.0663	44.0	7348	0.5416	0.8287	0.2685	1.2308	0.8287	0.8295	0.0883	0.0647
0.0652	45.0	7515	0.5431	0.8275	0.2697	1.2374	0.8275	0.8282	0.0932	0.0648
0.0652	46.0	7682	0.5433	0.8295	0.2693	1.2347	0.8295	0.8303	0.0891	0.0681
0.0652	47.0	7849	0.5441	0.8277	0.2696	1.2433	0.8277	0.8286	0.0882	0.0681
0.0651	48.0	8016	0.5439	0.8293	0.2695	1.2358	0.8293	0.8301	0.0888	0.0692
0.0651	49.0	8183	0.5445	0.8287	0.2696	1.2499	0.8287	0.8296	0.0882	0.0695
0.0651	50.0	8350	0.5443	0.8293	0.2695	1.2500	0.8293	0.8301	0.0874	0.0693

Framework versions

Transformers 4.26.1
Pytorch 1.13.1.post200
Datasets 2.9.0
Tokenizers 0.13.2

jordyvl/dit-base-finetuned-rvlcdip-small_rvl_cdip-NK1000_kd_CEKD_t2.5_a0.5

dit-base-finetuned-rvlcdip-small_rvl_cdip-NK1000_kd_CEKD_t2.5_a0.5

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions