vit-small_tobacco3482_simkd_CEKD_tNone_aNone_tNone_gNone

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: 0.0379
Accuracy: 0.8
Brier Loss: 0.6938
Nll: 1.3290
F1 Micro: 0.8000
F1 Macro: 0.7859
Ece: 0.5869
Aurc: 0.0931

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	0.0506	0.09	0.8991	6.5155	0.09	0.0484	0.1622	0.8986
No log	2.0	50	0.0468	0.22	0.8982	4.6950	0.22	0.1025	0.2491	0.7656
No log	3.0	75	0.0463	0.29	0.8969	3.3099	0.29	0.1676	0.2924	0.6888
No log	4.0	100	0.0459	0.37	0.8954	3.2920	0.37	0.1891	0.3517	0.4208
No log	5.0	125	0.0455	0.395	0.8929	3.2550	0.395	0.2299	0.3759	0.3617
No log	6.0	150	0.0449	0.49	0.8885	2.9109	0.49	0.3135	0.4396	0.2804
No log	7.0	175	0.0441	0.495	0.8796	2.8950	0.495	0.3248	0.4360	0.2721
No log	8.0	200	0.0430	0.545	0.8619	2.5199	0.545	0.3771	0.4777	0.2129
No log	9.0	225	0.0418	0.62	0.8382	2.2126	0.62	0.4291	0.5298	0.1659
No log	10.0	250	0.0409	0.645	0.8137	2.2525	0.645	0.4947	0.5293	0.1552
No log	11.0	275	0.0401	0.68	0.7863	2.4423	0.68	0.5145	0.5433	0.1215
No log	12.0	300	0.0392	0.68	0.7628	1.9779	0.68	0.5373	0.5402	0.1172
No log	13.0	325	0.0385	0.745	0.7350	1.8986	0.745	0.6126	0.5806	0.0843
No log	14.0	350	0.0384	0.735	0.7268	1.9922	0.735	0.6451	0.5466	0.0997
No log	15.0	375	0.0381	0.745	0.7180	1.6965	0.745	0.6627	0.5586	0.0761
No log	16.0	400	0.0377	0.805	0.7031	1.2564	0.805	0.7353	0.6034	0.0713
No log	17.0	425	0.0389	0.745	0.7303	1.5063	0.745	0.7192	0.5779	0.0705
No log	18.0	450	0.0387	0.765	0.7219	1.5776	0.765	0.7703	0.5815	0.0923
No log	19.0	475	0.0383	0.805	0.7213	1.3953	0.805	0.7906	0.6159	0.0667
0.0432	20.0	500	0.0377	0.835	0.6952	1.3075	0.835	0.8271	0.6116	0.0799
0.0432	21.0	525	0.0381	0.795	0.7018	1.6184	0.795	0.7723	0.5851	0.0880
0.0432	22.0	550	0.0378	0.81	0.6984	1.4292	0.81	0.7950	0.6103	0.0673
0.0432	23.0	575	0.0380	0.805	0.6976	1.4852	0.805	0.7951	0.5942	0.0808
0.0432	24.0	600	0.0377	0.825	0.6907	1.4501	0.825	0.8103	0.6020	0.0774
0.0432	25.0	625	0.0377	0.83	0.6920	1.4509	0.83	0.8148	0.6038	0.0759
0.0432	26.0	650	0.0377	0.825	0.6927	1.4113	0.825	0.8114	0.6072	0.0765
0.0432	27.0	675	0.0377	0.825	0.6924	1.4044	0.825	0.8114	0.6057	0.0757
0.0432	28.0	700	0.0377	0.82	0.6932	1.4521	0.82	0.8061	0.6017	0.0815
0.0432	29.0	725	0.0377	0.82	0.6932	1.3593	0.82	0.8080	0.5983	0.0794
0.0432	30.0	750	0.0377	0.82	0.6926	1.3437	0.82	0.8069	0.6042	0.0819
0.0432	31.0	775	0.0377	0.815	0.6932	1.3453	0.815	0.8027	0.5988	0.0815
0.0432	32.0	800	0.0377	0.82	0.6930	1.3384	0.82	0.8029	0.6044	0.0855
0.0432	33.0	825	0.0377	0.81	0.6928	1.3969	0.81	0.7927	0.5929	0.0835
0.0432	34.0	850	0.0378	0.805	0.6927	1.3995	0.805	0.7886	0.5961	0.0855
0.0432	35.0	875	0.0377	0.81	0.6927	1.3705	0.81	0.7979	0.5910	0.0887
0.0432	36.0	900	0.0378	0.805	0.6930	1.3566	0.805	0.7886	0.5850	0.0817
0.0432	37.0	925	0.0377	0.82	0.6927	1.3537	0.82	0.8022	0.5936	0.0847
0.0432	38.0	950	0.0377	0.815	0.6930	1.3574	0.815	0.7978	0.5976	0.0854
0.0432	39.0	975	0.0377	0.815	0.6932	1.4599	0.815	0.7978	0.5955	0.0864
0.035	40.0	1000	0.0377	0.815	0.6926	1.4147	0.815	0.7978	0.5990	0.0869
0.035	41.0	1025	0.0377	0.81	0.6931	1.4065	0.81	0.7943	0.5966	0.0844
0.035	42.0	1050	0.0378	0.81	0.6929	1.4678	0.81	0.7961	0.5902	0.0891
0.035	43.0	1075	0.0378	0.81	0.6927	1.4164	0.81	0.7971	0.5951	0.0897
0.035	44.0	1100	0.0378	0.81	0.6930	1.4646	0.81	0.7961	0.5948	0.0875
0.035	45.0	1125	0.0378	0.815	0.6921	1.4660	0.815	0.8004	0.6024	0.0895
0.035	46.0	1150	0.0378	0.81	0.6929	1.4098	0.81	0.7961	0.5987	0.0831
0.035	47.0	1175	0.0378	0.815	0.6928	1.4634	0.815	0.8004	0.5963	0.0911
0.035	48.0	1200	0.0378	0.81	0.6932	1.4648	0.81	0.7961	0.5841	0.0877
0.035	49.0	1225	0.0378	0.81	0.6928	1.4635	0.81	0.7961	0.5955	0.0898
0.035	50.0	1250	0.0378	0.805	0.6935	1.4688	0.805	0.7882	0.5795	0.0902
0.035	51.0	1275	0.0378	0.805	0.6928	1.4665	0.805	0.7882	0.5848	0.0916
0.035	52.0	1300	0.0378	0.81	0.6925	1.4249	0.81	0.7961	0.5869	0.0926
0.035	53.0	1325	0.0378	0.815	0.6926	1.4150	0.815	0.8021	0.5934	0.0913
0.035	54.0	1350	0.0378	0.81	0.6929	1.4155	0.81	0.7961	0.5943	0.0913
0.035	55.0	1375	0.0378	0.805	0.6928	1.4141	0.805	0.7882	0.5934	0.0964
0.035	56.0	1400	0.0378	0.805	0.6930	1.4124	0.805	0.7882	0.5926	0.0958
0.035	57.0	1425	0.0378	0.81	0.6935	1.4116	0.81	0.7934	0.6002	0.0895
0.035	58.0	1450	0.0378	0.805	0.6928	1.4059	0.805	0.7882	0.5890	0.0937
0.035	59.0	1475	0.0378	0.805	0.6929	1.4141	0.805	0.7882	0.5918	0.0967
0.0348	60.0	1500	0.0378	0.81	0.6935	1.4086	0.81	0.7934	0.5915	0.0934
0.0348	61.0	1525	0.0378	0.81	0.6930	1.4105	0.81	0.7941	0.5954	0.0961
0.0348	62.0	1550	0.0378	0.81	0.6933	1.4166	0.81	0.7941	0.5889	0.0954
0.0348	63.0	1575	0.0378	0.81	0.6933	1.4109	0.81	0.7934	0.5963	0.0975
0.0348	64.0	1600	0.0378	0.81	0.6932	1.4131	0.81	0.7934	0.5980	0.0953
0.0348	65.0	1625	0.0378	0.81	0.6937	1.4182	0.81	0.7934	0.5956	0.0970
0.0348	66.0	1650	0.0378	0.805	0.6933	1.4125	0.805	0.7893	0.5905	0.0966
0.0348	67.0	1675	0.0378	0.81	0.6937	1.4136	0.81	0.7934	0.5965	0.0975
0.0348	68.0	1700	0.0379	0.81	0.6935	1.4137	0.81	0.7934	0.5994	0.0971
0.0348	69.0	1725	0.0378	0.805	0.6935	1.4196	0.805	0.7893	0.5913	0.0946
0.0348	70.0	1750	0.0379	0.805	0.6933	1.4129	0.805	0.7893	0.5877	0.0945
0.0348	71.0	1775	0.0379	0.805	0.6933	1.4172	0.805	0.7893	0.5921	0.0951
0.0348	72.0	1800	0.0379	0.805	0.6931	1.4136	0.805	0.7893	0.5851	0.0953
0.0348	73.0	1825	0.0379	0.805	0.6929	1.4168	0.805	0.7893	0.5846	0.0971
0.0348	74.0	1850	0.0379	0.805	0.6939	1.4185	0.805	0.7893	0.5892	0.0950
0.0348	75.0	1875	0.0379	0.805	0.6935	1.4171	0.805	0.7893	0.5946	0.0938
0.0348	76.0	1900	0.0379	0.805	0.6934	1.4217	0.805	0.7893	0.5939	0.0959
0.0348	77.0	1925	0.0379	0.8	0.6932	1.4162	0.8000	0.7859	0.5826	0.0954
0.0348	78.0	1950	0.0379	0.8	0.6935	1.4172	0.8000	0.7859	0.5912	0.0950
0.0348	79.0	1975	0.0379	0.8	0.6933	1.4169	0.8000	0.7859	0.5885	0.0964
0.0348	80.0	2000	0.0379	0.8	0.6935	1.4196	0.8000	0.7859	0.5865	0.0957
0.0348	81.0	2025	0.0379	0.8	0.6937	1.4213	0.8000	0.7859	0.5880	0.0962
0.0348	82.0	2050	0.0379	0.8	0.6939	1.4201	0.8000	0.7859	0.5910	0.0962
0.0348	83.0	2075	0.0379	0.8	0.6938	1.3762	0.8000	0.7859	0.5883	0.0945
0.0348	84.0	2100	0.0379	0.8	0.6938	1.4218	0.8000	0.7859	0.5947	0.0950
0.0348	85.0	2125	0.0379	0.8	0.6935	1.3657	0.8000	0.7859	0.5857	0.0912
0.0348	86.0	2150	0.0379	0.8	0.6938	1.3278	0.8000	0.7859	0.5892	0.0929
0.0348	87.0	2175	0.0379	0.8	0.6938	1.3831	0.8000	0.7859	0.5856	0.0946
0.0348	88.0	2200	0.0379	0.8	0.6938	1.3761	0.8000	0.7859	0.5892	0.0955
0.0348	89.0	2225	0.0379	0.8	0.6938	1.3296	0.8000	0.7859	0.5870	0.0947
0.0348	90.0	2250	0.0379	0.8	0.6939	1.3667	0.8000	0.7859	0.5909	0.0926
0.0348	91.0	2275	0.0379	0.8	0.6940	1.3346	0.8000	0.7859	0.5906	0.0930
0.0348	92.0	2300	0.0379	0.8	0.6938	1.3268	0.8000	0.7859	0.5870	0.0936
0.0348	93.0	2325	0.0379	0.8	0.6937	1.3320	0.8000	0.7859	0.5919	0.0939
0.0348	94.0	2350	0.0379	0.8	0.6939	1.3324	0.8000	0.7859	0.5870	0.0928
0.0348	95.0	2375	0.0379	0.8	0.6937	1.3289	0.8000	0.7859	0.5869	0.0932
0.0348	96.0	2400	0.0379	0.8	0.6938	1.3264	0.8000	0.7859	0.5869	0.0931
0.0348	97.0	2425	0.0379	0.8	0.6938	1.3280	0.8000	0.7859	0.5870	0.0932
0.0348	98.0	2450	0.0379	0.8	0.6938	1.3297	0.8000	0.7859	0.5869	0.0930
0.0348	99.0	2475	0.0379	0.8	0.6938	1.3304	0.8000	0.7859	0.5869	0.0929
0.0347	100.0	2500	0.0379	0.8	0.6938	1.3290	0.8000	0.7859	0.5869	0.0931

Framework versions

Transformers 4.26.1
Pytorch 1.13.1.post200
Datasets 2.9.0
Tokenizers 0.13.2

jordyvl/vit-small_tobacco3482_simkd_CEKD_tNone_aNone_tNone_gNone

vit-small_tobacco3482_simkd_CEKD_tNone_aNone_tNone_gNone

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions