vit-tiny_tobacco3482_simkd_CEKD_tNone_aNone_tNone_gNone

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: 0.0396
Accuracy: 0.735
Brier Loss: 0.7729
Nll: 1.4473
F1 Micro: 0.735
F1 Macro: 0.6948
Ece: 0.5886
Aurc: 0.0947

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: 32
eval_batch_size: 32
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: 100

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Brier Loss	Nll	F1 Micro	F1 Macro	Ece	Aurc
No log	1.0	25	0.0553	0.085	0.8991	5.2518	0.085	0.0595	0.1614	0.8792
No log	2.0	50	0.0488	0.035	0.9007	7.4288	0.035	0.0069	0.1218	0.9410
No log	3.0	75	0.0481	0.045	0.8999	6.0525	0.045	0.0087	0.1349	0.9308
No log	4.0	100	0.0478	0.05	0.8991	5.6444	0.0500	0.0149	0.1378	0.9211
No log	5.0	125	0.0475	0.14	0.8981	5.8239	0.14	0.0863	0.1987	0.8452
No log	6.0	150	0.0471	0.305	0.8964	5.7469	0.305	0.1652	0.3097	0.5016
No log	7.0	175	0.0466	0.305	0.8950	4.9568	0.305	0.1899	0.3035	0.5165
No log	8.0	200	0.0461	0.315	0.8931	4.8214	0.315	0.1811	0.3152	0.4687
No log	9.0	225	0.0455	0.315	0.8907	4.7406	0.315	0.2028	0.3225	0.4671
No log	10.0	250	0.0449	0.35	0.8862	4.7538	0.35	0.1972	0.3364	0.4263
No log	11.0	275	0.0443	0.37	0.8793	4.6283	0.37	0.2106	0.3455	0.4084
No log	12.0	300	0.0438	0.4	0.8731	3.9664	0.4000	0.2443	0.3685	0.3731
No log	13.0	325	0.0434	0.425	0.8628	3.9702	0.425	0.2574	0.3842	0.3601
No log	14.0	350	0.0430	0.465	0.8586	3.8630	0.465	0.3226	0.4112	0.3024
No log	15.0	375	0.0428	0.46	0.8488	3.9046	0.46	0.2693	0.4082	0.2854
No log	16.0	400	0.0424	0.475	0.8430	3.2916	0.4750	0.2802	0.4183	0.2626
No log	17.0	425	0.0421	0.555	0.8439	2.7780	0.555	0.4109	0.4760	0.2123
No log	18.0	450	0.0418	0.575	0.8317	2.8629	0.575	0.4399	0.4869	0.2123
No log	19.0	475	0.0415	0.665	0.8329	2.5145	0.665	0.5077	0.5655	0.1361
0.0491	20.0	500	0.0412	0.635	0.8121	2.7489	0.635	0.5155	0.5235	0.1686
0.0491	21.0	525	0.0410	0.655	0.8221	1.7853	0.655	0.5182	0.5509	0.1545
0.0491	22.0	550	0.0406	0.685	0.8045	1.5894	0.685	0.5486	0.5627	0.1305
0.0491	23.0	575	0.0405	0.68	0.7984	1.7241	0.68	0.5489	0.5545	0.1296
0.0491	24.0	600	0.0402	0.725	0.7959	1.5667	0.7250	0.6156	0.5926	0.1055
0.0491	25.0	625	0.0402	0.68	0.7927	1.4334	0.68	0.5853	0.5453	0.1239
0.0491	26.0	650	0.0401	0.705	0.7808	1.8114	0.705	0.5856	0.5735	0.1109
0.0491	27.0	675	0.0399	0.71	0.7859	1.6101	0.7100	0.6176	0.5679	0.1034
0.0491	28.0	700	0.0399	0.715	0.7808	1.3423	0.715	0.6612	0.5582	0.1130
0.0491	29.0	725	0.0398	0.705	0.7789	1.3921	0.705	0.6477	0.5615	0.1175
0.0491	30.0	750	0.0397	0.73	0.7767	1.5801	0.7300	0.6758	0.5741	0.1069
0.0491	31.0	775	0.0397	0.72	0.7774	1.3193	0.72	0.6653	0.5790	0.1004
0.0491	32.0	800	0.0396	0.745	0.7729	1.4864	0.745	0.6931	0.5941	0.0933
0.0491	33.0	825	0.0396	0.74	0.7736	1.5161	0.74	0.6901	0.5828	0.0934
0.0491	34.0	850	0.0396	0.745	0.7754	1.5432	0.745	0.6963	0.5911	0.0857
0.0491	35.0	875	0.0396	0.74	0.7744	1.4773	0.74	0.6936	0.5966	0.0896
0.0491	36.0	900	0.0397	0.715	0.7762	1.3769	0.715	0.6827	0.5675	0.1048
0.0491	37.0	925	0.0396	0.72	0.7744	1.3882	0.72	0.6780	0.5689	0.0970
0.0491	38.0	950	0.0396	0.72	0.7762	1.4098	0.72	0.6874	0.5701	0.1016
0.0491	39.0	975	0.0395	0.74	0.7728	1.3890	0.74	0.6894	0.5861	0.0902
0.0386	40.0	1000	0.0396	0.74	0.7724	1.5265	0.74	0.6936	0.5906	0.0881
0.0386	41.0	1025	0.0396	0.725	0.7730	1.3516	0.7250	0.6768	0.5784	0.0942
0.0386	42.0	1050	0.0396	0.73	0.7728	1.3633	0.7300	0.6847	0.5899	0.0945
0.0386	43.0	1075	0.0396	0.735	0.7730	1.3670	0.735	0.6874	0.5830	0.0940
0.0386	44.0	1100	0.0395	0.73	0.7727	1.4707	0.7300	0.6850	0.5914	0.0930
0.0386	45.0	1125	0.0396	0.725	0.7721	1.4269	0.7250	0.6810	0.5770	0.0934
0.0386	46.0	1150	0.0396	0.72	0.7730	1.3567	0.72	0.6793	0.5717	0.0976
0.0386	47.0	1175	0.0396	0.715	0.7731	1.3708	0.715	0.6757	0.5717	0.0974
0.0386	48.0	1200	0.0396	0.735	0.7724	1.4118	0.735	0.6874	0.5791	0.0923
0.0386	49.0	1225	0.0396	0.72	0.7729	1.3647	0.72	0.6837	0.5711	0.0965
0.0386	50.0	1250	0.0396	0.725	0.7727	1.3773	0.7250	0.6820	0.5740	0.0963
0.0386	51.0	1275	0.0396	0.73	0.7736	1.3286	0.7300	0.6847	0.5766	0.0939
0.0386	52.0	1300	0.0396	0.725	0.7732	1.3810	0.7250	0.6817	0.5830	0.0944
0.0386	53.0	1325	0.0396	0.725	0.7725	1.3568	0.7250	0.6820	0.5763	0.0948
0.0386	54.0	1350	0.0396	0.73	0.7731	1.3693	0.7300	0.6847	0.5768	0.0941
0.0386	55.0	1375	0.0396	0.745	0.7728	1.3631	0.745	0.7112	0.5842	0.0928
0.0386	56.0	1400	0.0396	0.715	0.7731	1.4175	0.715	0.6712	0.5600	0.0976
0.0386	57.0	1425	0.0396	0.725	0.7725	1.3668	0.7250	0.6929	0.5738	0.0962
0.0386	58.0	1450	0.0396	0.73	0.7734	1.3903	0.7300	0.6958	0.5868	0.0963
0.0386	59.0	1475	0.0396	0.725	0.7729	1.4120	0.7250	0.6765	0.5756	0.0945
0.0373	60.0	1500	0.0396	0.725	0.7732	1.3655	0.7250	0.6820	0.5754	0.0951
0.0373	61.0	1525	0.0396	0.745	0.7727	1.3676	0.745	0.7038	0.5913	0.0921
0.0373	62.0	1550	0.0396	0.72	0.7729	1.3629	0.72	0.6797	0.5762	0.0969
0.0373	63.0	1575	0.0396	0.725	0.7730	1.4242	0.7250	0.6865	0.5811	0.0950
0.0373	64.0	1600	0.0396	0.725	0.7735	1.3658	0.7250	0.6923	0.5750	0.0959
0.0373	65.0	1625	0.0396	0.73	0.7731	1.4296	0.7300	0.6958	0.5769	0.0954
0.0373	66.0	1650	0.0396	0.735	0.7727	1.4780	0.735	0.6980	0.5851	0.0938
0.0373	67.0	1675	0.0396	0.725	0.7725	1.3669	0.7250	0.6824	0.5715	0.0938
0.0373	68.0	1700	0.0396	0.725	0.7730	1.4327	0.7250	0.6804	0.5741	0.0940
0.0373	69.0	1725	0.0396	0.73	0.7728	1.3811	0.7300	0.6961	0.5806	0.0963
0.0373	70.0	1750	0.0396	0.735	0.7727	1.3812	0.735	0.7081	0.5765	0.0952
0.0373	71.0	1775	0.0396	0.73	0.7730	1.4263	0.7300	0.6961	0.5739	0.0953
0.0373	72.0	1800	0.0396	0.73	0.7731	1.4280	0.7300	0.6953	0.5803	0.0956
0.0373	73.0	1825	0.0396	0.735	0.7729	1.3676	0.735	0.6988	0.5889	0.0953
0.0373	74.0	1850	0.0396	0.735	0.7727	1.4358	0.735	0.6985	0.5828	0.0940
0.0373	75.0	1875	0.0396	0.735	0.7727	1.4306	0.735	0.6965	0.5786	0.0940
0.0373	76.0	1900	0.0396	0.73	0.7729	1.4343	0.7300	0.6957	0.5802	0.0958
0.0373	77.0	1925	0.0396	0.73	0.7726	1.4259	0.7300	0.6961	0.5795	0.0962
0.0373	78.0	1950	0.0396	0.74	0.7731	1.4246	0.74	0.7080	0.5879	0.0941
0.0373	79.0	1975	0.0396	0.735	0.7730	1.4414	0.735	0.6980	0.5914	0.0945
0.0372	80.0	2000	0.0396	0.74	0.7727	1.4285	0.74	0.7103	0.5915	0.0939
0.0372	81.0	2025	0.0396	0.735	0.7731	1.4379	0.735	0.6980	0.5826	0.0942
0.0372	82.0	2050	0.0396	0.735	0.7729	1.4308	0.735	0.6963	0.5827	0.0942
0.0372	83.0	2075	0.0396	0.735	0.7728	1.4329	0.735	0.6968	0.5896	0.0946
0.0372	84.0	2100	0.0396	0.735	0.7728	1.4343	0.735	0.6948	0.5889	0.0947
0.0372	85.0	2125	0.0396	0.735	0.7727	1.4320	0.735	0.6948	0.5988	0.0945
0.0372	86.0	2150	0.0396	0.735	0.7730	1.4366	0.735	0.6963	0.5883	0.0949
0.0372	87.0	2175	0.0396	0.73	0.7728	1.4825	0.7300	0.6888	0.5878	0.0945
0.0372	88.0	2200	0.0396	0.735	0.7731	1.4339	0.735	0.6945	0.5828	0.0948
0.0372	89.0	2225	0.0396	0.735	0.7729	1.4383	0.735	0.6948	0.5917	0.0946
0.0372	90.0	2250	0.0396	0.735	0.7729	1.4471	0.735	0.6948	0.5867	0.0944
0.0372	91.0	2275	0.0396	0.735	0.7728	1.4402	0.735	0.6948	0.5892	0.0946
0.0372	92.0	2300	0.0396	0.735	0.7729	1.4412	0.735	0.6948	0.5952	0.0948
0.0372	93.0	2325	0.0396	0.735	0.7729	1.4709	0.735	0.6948	0.5917	0.0948
0.0372	94.0	2350	0.0396	0.735	0.7728	1.4413	0.735	0.6948	0.5858	0.0947
0.0372	95.0	2375	0.0396	0.735	0.7729	1.4422	0.735	0.6948	0.5917	0.0946
0.0372	96.0	2400	0.0396	0.735	0.7729	1.4527	0.735	0.6948	0.5917	0.0946
0.0372	97.0	2425	0.0396	0.735	0.7729	1.4441	0.735	0.6948	0.5917	0.0946
0.0372	98.0	2450	0.0396	0.735	0.7729	1.4423	0.735	0.6948	0.5917	0.0946
0.0372	99.0	2475	0.0396	0.735	0.7729	1.4457	0.735	0.6948	0.5886	0.0948
0.0372	100.0	2500	0.0396	0.735	0.7729	1.4473	0.735	0.6948	0.5886	0.0947

Framework versions

Transformers 4.26.1
Pytorch 1.13.1.post200
Datasets 2.9.0
Tokenizers 0.13.2

jordyvl/vit-tiny_tobacco3482_simkd_CEKD_tNone_aNone_tNone_gNone

vit-tiny_tobacco3482_simkd_CEKD_tNone_aNone_tNone_gNone

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions