dit-base-finetuned-rvlcdip-tiny_rvl_cdip-NK1000_kd

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: 0.5815
Accuracy: 0.8055
Brier Loss: 0.2836
Nll: 1.6135
F1 Micro: 0.8055
F1 Macro: 0.8061
Ece: 0.0597
Aurc: 0.0526

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	1.2844	0.5403	0.5889	3.0582	0.5403	0.5275	0.0742	0.2209
No log	2.0	250	0.9687	0.655	0.4587	2.4358	0.655	0.6414	0.0559	0.1296
No log	3.0	375	0.8401	0.7063	0.4019	2.2308	0.7063	0.7008	0.0588	0.0990
1.234	4.0	500	0.8080	0.7145	0.3874	2.1628	0.7145	0.7163	0.0487	0.0951
1.234	5.0	625	0.7772	0.7238	0.3755	2.0380	0.7237	0.7167	0.0421	0.0914
1.234	6.0	750	0.7530	0.7498	0.3484	2.1346	0.7498	0.7464	0.0477	0.0774
1.234	7.0	875	0.7034	0.7652	0.3267	2.0596	0.7652	0.7664	0.0467	0.0678
0.3976	8.0	1000	0.7390	0.7715	0.3350	2.0568	0.7715	0.7704	0.0448	0.0763
0.3976	9.0	1125	0.7019	0.7762	0.3209	2.0168	0.7762	0.7768	0.0556	0.0769
0.3976	10.0	1250	0.7318	0.7668	0.3346	2.1148	0.7668	0.7699	0.0529	0.0792
0.3976	11.0	1375	0.7083	0.7782	0.3213	2.0671	0.7782	0.7775	0.0452	0.0756
0.1591	12.0	1500	0.7535	0.7668	0.3424	2.1407	0.7668	0.7636	0.0564	0.0845
0.1591	13.0	1625	0.7117	0.775	0.3288	2.0935	0.775	0.7766	0.0525	0.0785
0.1591	14.0	1750	0.6421	0.785	0.3039	1.9939	0.785	0.7860	0.0512	0.0643
0.1591	15.0	1875	0.6475	0.7865	0.3050	1.9301	0.7865	0.7867	0.0552	0.0636
0.1125	16.0	2000	0.6477	0.7893	0.3064	1.9442	0.7893	0.7920	0.0556	0.0684
0.1125	17.0	2125	0.6509	0.7883	0.3113	1.8957	0.7883	0.7907	0.0498	0.0710
0.1125	18.0	2250	0.6291	0.7925	0.3038	1.8697	0.7925	0.7963	0.0512	0.0677
0.1125	19.0	2375	0.6279	0.7963	0.2992	1.8155	0.7963	0.7950	0.0478	0.0647
0.095	20.0	2500	0.6246	0.7937	0.3008	1.7925	0.7937	0.7946	0.0595	0.0659
0.095	21.0	2625	0.6149	0.7953	0.2962	1.8237	0.7953	0.7951	0.0547	0.0590
0.095	22.0	2750	0.6196	0.7953	0.3000	1.8031	0.7953	0.7969	0.0567	0.0643
0.095	23.0	2875	0.6023	0.798	0.2932	1.7663	0.798	0.7983	0.0497	0.0616
0.0829	24.0	3000	0.6107	0.7943	0.2951	1.7755	0.7943	0.7958	0.0564	0.0581
0.0829	25.0	3125	0.5986	0.8015	0.2930	1.7243	0.8015	0.8027	0.0565	0.0574
0.0829	26.0	3250	0.5899	0.8005	0.2886	1.7304	0.8005	0.8021	0.0546	0.0560
0.0829	27.0	3375	0.5836	0.8023	0.2846	1.6865	0.8023	0.8024	0.0479	0.0561
0.074	28.0	3500	0.5824	0.8047	0.2850	1.6817	0.8047	0.8060	0.0524	0.0559
0.074	29.0	3625	0.5760	0.8063	0.2822	1.6505	0.8062	0.8065	0.0500	0.0546
0.074	30.0	3750	0.5819	0.8065	0.2843	1.6667	0.8065	0.8079	0.0563	0.0544
0.074	31.0	3875	0.5800	0.8045	0.2841	1.6658	0.8045	0.8059	0.0511	0.0548
0.0668	32.0	4000	0.5828	0.8053	0.2841	1.6883	0.8053	0.8054	0.0559	0.0547
0.0668	33.0	4125	0.5802	0.8037	0.2838	1.6669	0.8037	0.8038	0.0572	0.0545
0.0668	34.0	4250	0.5772	0.8067	0.2821	1.6588	0.8067	0.8083	0.0520	0.0525
0.0668	35.0	4375	0.5745	0.807	0.2812	1.6524	0.807	0.8072	0.0528	0.0528
0.0631	36.0	4500	0.5770	0.8063	0.2826	1.6433	0.8062	0.8071	0.0559	0.0528
0.0631	37.0	4625	0.5782	0.8007	0.2837	1.5953	0.8007	0.8021	0.0581	0.0541
0.0631	38.0	4750	0.5780	0.8047	0.2829	1.6275	0.8047	0.8052	0.0540	0.0521
0.0631	39.0	4875	0.5759	0.8055	0.2817	1.6162	0.8055	0.8065	0.0528	0.0529
0.0612	40.0	5000	0.5770	0.8047	0.2825	1.6131	0.8047	0.8051	0.0575	0.0524
0.0612	41.0	5125	0.5771	0.8043	0.2819	1.6015	0.8043	0.8048	0.0562	0.0519
0.0612	42.0	5250	0.5776	0.8043	0.2825	1.6152	0.8043	0.8047	0.0566	0.0527
0.0612	43.0	5375	0.5793	0.8057	0.2830	1.6196	0.8057	0.8065	0.0538	0.0527
0.06	44.0	5500	0.5801	0.8053	0.2835	1.6183	0.8053	0.8060	0.0618	0.0527
0.06	45.0	5625	0.5800	0.805	0.2831	1.6057	0.805	0.8055	0.0568	0.0530
0.06	46.0	5750	0.5812	0.805	0.2836	1.6034	0.805	0.8056	0.0577	0.0529
0.06	47.0	5875	0.5809	0.805	0.2834	1.6164	0.805	0.8056	0.0580	0.0526
0.0593	48.0	6000	0.5810	0.8057	0.2834	1.6108	0.8057	0.8064	0.0617	0.0525
0.0593	49.0	6125	0.5812	0.8053	0.2836	1.6140	0.8053	0.8058	0.0570	0.0527
0.0593	50.0	6250	0.5815	0.8055	0.2836	1.6135	0.8055	0.8061	0.0597	0.0526

Framework versions

Transformers 4.26.1
Pytorch 1.13.1.post200
Datasets 2.9.0
Tokenizers 0.13.2

jordyvl/dit-base-finetuned-rvlcdip-tiny_rvl_cdip-NK1000_kd

dit-base-finetuned-rvlcdip-tiny_rvl_cdip-NK1000_kd

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions