dit-base-tiny_rvl_cdip-NK1000_hint

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: 2.5793
Accuracy: 0.8185
Brier Loss: 0.3293
Nll: 2.0402
F1 Micro: 0.8185
F1 Macro: 0.8184
Ece: 0.1589
Aurc: 0.0584

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: 16
eval_batch_size: 16
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
2.7601	1.0	1000	2.6212	0.5952	0.5281	2.4218	0.5952	0.5899	0.0529	0.1801
2.236	2.0	2000	2.1744	0.6847	0.4219	2.1650	0.6847	0.6827	0.0515	0.1135
1.9525	3.0	3000	1.9941	0.7147	0.3889	2.0860	0.7147	0.7163	0.0616	0.0961
1.729	4.0	4000	1.8880	0.7485	0.3585	2.1097	0.7485	0.7476	0.0649	0.0811
1.6172	5.0	5000	1.8693	0.7455	0.3598	2.0514	0.7455	0.7448	0.0733	0.0811
1.4385	6.0	6000	1.8350	0.747	0.3654	2.0917	0.747	0.7536	0.0828	0.0843
1.3098	7.0	7000	1.8317	0.7625	0.3515	2.0367	0.7625	0.7668	0.1039	0.0794
1.1854	8.0	8000	1.8872	0.7675	0.3589	2.1085	0.7675	0.7661	0.1294	0.0694
1.1066	9.0	9000	1.9843	0.7642	0.3707	2.1492	0.7642	0.7670	0.1457	0.0715
1.0518	10.0	10000	1.9993	0.77	0.3660	2.0694	0.7700	0.7702	0.1440	0.0695
0.9741	11.0	11000	2.2346	0.769	0.3870	2.0588	0.769	0.7704	0.1748	0.0735
0.938	12.0	12000	2.2626	0.767	0.3918	2.1412	0.767	0.7679	0.1774	0.0723
0.899	13.0	13000	2.3671	0.7698	0.3995	2.1064	0.7698	0.7725	0.1820	0.0708
0.8708	14.0	14000	2.3386	0.7768	0.3838	2.1948	0.7768	0.7777	0.1724	0.0683
0.8456	15.0	15000	2.3234	0.7827	0.3774	2.0937	0.7828	0.7830	0.1752	0.0638
0.8021	16.0	16000	2.5700	0.7685	0.4092	2.1458	0.7685	0.7684	0.1911	0.0717
0.7921	17.0	17000	2.5721	0.778	0.3934	2.1774	0.778	0.7756	0.1829	0.0708
0.7779	18.0	18000	2.7204	0.772	0.4089	2.1145	0.772	0.7707	0.1926	0.0758
0.7484	19.0	19000	2.7208	0.7752	0.4044	2.1020	0.7752	0.7760	0.1901	0.0746
0.7469	20.0	20000	2.5898	0.7927	0.3711	2.2076	0.7927	0.7909	0.1763	0.0648
0.7256	21.0	21000	2.5658	0.791	0.3727	2.0710	0.791	0.7920	0.1763	0.0650
0.7137	22.0	22000	2.6782	0.7847	0.3851	2.1469	0.7847	0.7854	0.1824	0.0703
0.6912	23.0	23000	2.5574	0.802	0.3539	2.1340	0.802	0.8009	0.1685	0.0615
0.6894	24.0	24000	2.6331	0.7913	0.3760	2.0913	0.7913	0.7944	0.1807	0.0674
0.6692	25.0	25000	2.6074	0.7955	0.3658	2.0837	0.7955	0.7975	0.1745	0.0645
0.6541	26.0	26000	2.6059	0.7945	0.3672	2.0798	0.7945	0.7936	0.1751	0.0616
0.6517	27.0	27000	2.7149	0.7965	0.3697	2.0842	0.7965	0.7987	0.1769	0.0663
0.6484	28.0	28000	2.5700	0.8047	0.3542	2.0142	0.8047	0.8058	0.1685	0.0597
0.6342	29.0	29000	2.6774	0.7987	0.3660	2.1231	0.7987	0.7972	0.1759	0.0622
0.6331	30.0	30000	2.6112	0.7973	0.3621	2.0740	0.7973	0.7981	0.1752	0.0621
0.6204	31.0	31000	2.6470	0.807	0.3521	2.0337	0.807	0.8056	0.1683	0.0638
0.612	32.0	32000	2.6265	0.8053	0.3549	2.0800	0.8053	0.8038	0.1678	0.0596
0.6049	33.0	33000	2.5749	0.8107	0.3428	2.0235	0.8108	0.8114	0.1662	0.0576
0.5984	34.0	34000	2.6667	0.804	0.3572	2.1015	0.804	0.8041	0.1726	0.0632
0.5961	35.0	35000	2.6481	0.8067	0.3521	2.0652	0.8067	0.8068	0.1673	0.0610
0.5958	36.0	36000	2.6831	0.8065	0.3523	2.0272	0.8065	0.8074	0.1681	0.0623
0.5836	37.0	37000	2.6573	0.8067	0.3533	2.0149	0.8067	0.8050	0.1699	0.0624
0.5855	38.0	38000	2.6730	0.8087	0.3510	2.0149	0.8087	0.8087	0.1693	0.0611
0.581	39.0	39000	2.6464	0.815	0.3390	2.0609	0.815	0.8148	0.1628	0.0588
0.5708	40.0	40000	2.7165	0.8077	0.3515	2.0489	0.8077	0.8078	0.1698	0.0644
0.5727	41.0	41000	2.6264	0.8135	0.3402	2.0070	0.8135	0.8130	0.1643	0.0601
0.5688	42.0	42000	2.6522	0.8077	0.3522	2.0064	0.8077	0.8076	0.1687	0.0615
0.5648	43.0	43000	2.5806	0.8193	0.3295	2.0211	0.8193	0.8188	0.1593	0.0585
0.5649	44.0	44000	2.6182	0.8125	0.3372	2.0292	0.8125	0.8122	0.1630	0.0598
0.562	45.0	45000	2.6274	0.8157	0.3366	2.0047	0.8157	0.8158	0.1610	0.0577
0.5592	46.0	46000	2.6069	0.8145	0.3370	2.0276	0.8145	0.8146	0.1632	0.0599
0.5582	47.0	47000	2.5935	0.8187	0.3331	2.0291	0.8187	0.8193	0.1594	0.0588
0.5565	48.0	48000	2.5780	0.8197	0.3302	2.0401	0.8197	0.8197	0.1580	0.0577
0.5571	49.0	49000	2.5781	0.8185	0.3294	2.0367	0.8185	0.8184	0.1590	0.0583
0.5568	50.0	50000	2.5793	0.8185	0.3293	2.0402	0.8185	0.8184	0.1589	0.0584

Framework versions

Transformers 4.26.1
Pytorch 1.13.1.post200
Datasets 2.9.0
Tokenizers 0.13.2

jordyvl/dit-base-tiny_rvl_cdip-NK1000_hint

dit-base-tiny_rvl_cdip-NK1000_hint

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions