dit-base_tobacco-tiny_tobacco3482_simkd

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: 0.7298
Accuracy: 0.8
Brier Loss: 0.3356
Nll: 1.1950
F1 Micro: 0.8000
F1 Macro: 0.7677
Ece: 0.2868
Aurc: 0.0614

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: 100

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Brier Loss	Nll	F1 Micro	F1 Macro	Ece	Aurc
No log	1.0	50	1.0044	0.11	0.8970	5.3755	0.11	0.0297	0.1810	0.9082
No log	2.0	100	0.9997	0.27	0.8946	5.6759	0.27	0.1038	0.2752	0.7229
No log	3.0	150	0.9946	0.345	0.8902	4.6234	0.345	0.1969	0.3377	0.6577
No log	4.0	200	0.9814	0.4	0.8686	3.0912	0.4000	0.2605	0.3687	0.3808
No log	5.0	250	0.9618	0.56	0.8277	2.9065	0.56	0.4439	0.4769	0.2239
No log	6.0	300	0.9225	0.58	0.7429	2.5647	0.58	0.4408	0.4561	0.1944
No log	7.0	350	0.8843	0.705	0.6414	2.4145	0.705	0.5531	0.4493	0.1261
No log	8.0	400	0.8627	0.685	0.5773	2.4171	0.685	0.5755	0.3710	0.1378
No log	9.0	450	0.8252	0.73	0.5158	1.6133	0.7300	0.6403	0.3706	0.1066
0.9306	10.0	500	0.8164	0.74	0.4861	1.9299	0.74	0.6672	0.3352	0.1090
0.9306	11.0	550	0.8350	0.67	0.5078	2.0291	0.67	0.6083	0.3271	0.1514
0.9306	12.0	600	0.8089	0.695	0.4680	1.6726	0.695	0.6065	0.3049	0.1040
0.9306	13.0	650	0.7847	0.78	0.4097	1.3710	0.78	0.7067	0.3090	0.0825
0.9306	14.0	700	0.7793	0.8	0.3952	1.4382	0.8000	0.7351	0.3189	0.0823
0.9306	15.0	750	0.7756	0.775	0.3979	1.2640	0.775	0.6997	0.2950	0.0835
0.9306	16.0	800	0.7888	0.765	0.3927	1.2499	0.765	0.6894	0.3175	0.0719
0.9306	17.0	850	0.7596	0.795	0.3603	1.1834	0.795	0.7250	0.2930	0.0673
0.9306	18.0	900	0.7581	0.795	0.3580	1.1902	0.795	0.7241	0.3104	0.0665
0.9306	19.0	950	0.7546	0.81	0.3547	1.1055	0.81	0.7583	0.3024	0.0621
0.7329	20.0	1000	0.7520	0.81	0.3547	1.1284	0.81	0.7533	0.3209	0.0581
0.7329	21.0	1050	0.7669	0.775	0.3906	1.3812	0.775	0.7502	0.3212	0.0794
0.7329	22.0	1100	0.7532	0.81	0.3591	1.0982	0.81	0.7836	0.3035	0.0708
0.7329	23.0	1150	0.7519	0.805	0.3643	1.0628	0.805	0.7742	0.2813	0.0732
0.7329	24.0	1200	0.7494	0.795	0.3614	1.1123	0.795	0.7618	0.2988	0.0699
0.7329	25.0	1250	0.7517	0.79	0.3696	1.0703	0.79	0.7606	0.3081	0.0800
0.7329	26.0	1300	0.7513	0.795	0.3629	1.1020	0.795	0.7769	0.2797	0.0722
0.7329	27.0	1350	0.7485	0.795	0.3552	1.0352	0.795	0.7671	0.2678	0.0684
0.7329	28.0	1400	0.7442	0.805	0.3471	1.0956	0.805	0.7706	0.2807	0.0630
0.7329	29.0	1450	0.7473	0.795	0.3592	1.1204	0.795	0.7685	0.2897	0.0722
0.6917	30.0	1500	0.7449	0.815	0.3482	1.0584	0.815	0.7862	0.2949	0.0629
0.6917	31.0	1550	0.7443	0.8	0.3512	1.1010	0.8000	0.7770	0.2954	0.0622
0.6917	32.0	1600	0.7454	0.785	0.3543	1.0994	0.785	0.7631	0.2957	0.0639
0.6917	33.0	1650	0.7421	0.815	0.3449	1.1826	0.815	0.7853	0.2996	0.0592
0.6917	34.0	1700	0.7454	0.79	0.3559	1.1000	0.79	0.7597	0.2964	0.0659
0.6917	35.0	1750	0.7418	0.815	0.3477	1.1616	0.815	0.7867	0.3133	0.0617
0.6917	36.0	1800	0.7425	0.815	0.3464	1.1274	0.815	0.7949	0.3173	0.0578
0.6917	37.0	1850	0.7421	0.8	0.3448	1.1909	0.8000	0.7732	0.2900	0.0639
0.6917	38.0	1900	0.7415	0.795	0.3471	1.1816	0.795	0.7594	0.2860	0.0655
0.6917	39.0	1950	0.7405	0.78	0.3502	1.1084	0.78	0.7491	0.2709	0.0650
0.6764	40.0	2000	0.7398	0.81	0.3457	1.1746	0.81	0.7797	0.2973	0.0603
0.6764	41.0	2050	0.7394	0.805	0.3437	1.1201	0.805	0.7764	0.2915	0.0626
0.6764	42.0	2100	0.7380	0.81	0.3420	1.0987	0.81	0.7861	0.2815	0.0583
0.6764	43.0	2150	0.7386	0.8	0.3437	1.1855	0.8000	0.7667	0.2804	0.0617
0.6764	44.0	2200	0.7398	0.795	0.3437	1.1138	0.795	0.7660	0.2719	0.0614
0.6764	45.0	2250	0.7384	0.805	0.3441	1.1100	0.805	0.7699	0.3065	0.0628
0.6764	46.0	2300	0.7389	0.79	0.3488	1.1079	0.79	0.7552	0.2615	0.0647
0.6764	47.0	2350	0.7368	0.8	0.3440	1.1095	0.8000	0.7698	0.2908	0.0624
0.6764	48.0	2400	0.7365	0.8	0.3452	1.0995	0.8000	0.7739	0.2838	0.0645
0.6764	49.0	2450	0.7365	0.8	0.3367	1.0442	0.8000	0.7712	0.2735	0.0585
0.6662	50.0	2500	0.7342	0.815	0.3379	1.1009	0.815	0.7815	0.2964	0.0584
0.6662	51.0	2550	0.7340	0.805	0.3358	1.0985	0.805	0.7723	0.2635	0.0593
0.6662	52.0	2600	0.7370	0.8	0.3429	1.1227	0.8000	0.7709	0.2841	0.0603
0.6662	53.0	2650	0.7325	0.81	0.3380	1.1110	0.81	0.7790	0.3022	0.0601
0.6662	54.0	2700	0.7320	0.8	0.3363	1.0621	0.8000	0.7647	0.2815	0.0607
0.6662	55.0	2750	0.7324	0.805	0.3321	0.9926	0.805	0.7693	0.2972	0.0600
0.6662	56.0	2800	0.7318	0.805	0.3364	1.0537	0.805	0.7681	0.2554	0.0612
0.6662	57.0	2850	0.7311	0.82	0.3355	1.1133	0.82	0.7862	0.2776	0.0594
0.6662	58.0	2900	0.7317	0.81	0.3331	1.0662	0.81	0.7797	0.2600	0.0579
0.6662	59.0	2950	0.7327	0.805	0.3382	1.1876	0.805	0.7735	0.2797	0.0621
0.6577	60.0	3000	0.7322	0.8	0.3356	1.1864	0.8000	0.7680	0.2797	0.0612
0.6577	61.0	3050	0.7327	0.795	0.3391	1.1347	0.795	0.7614	0.2883	0.0641
0.6577	62.0	3100	0.7315	0.815	0.3364	1.1227	0.815	0.7848	0.2681	0.0599
0.6577	63.0	3150	0.7316	0.805	0.3392	1.0608	0.805	0.7717	0.2742	0.0632
0.6577	64.0	3200	0.7313	0.82	0.3341	1.0601	0.82	0.7878	0.2950	0.0583
0.6577	65.0	3250	0.7322	0.805	0.3388	1.1837	0.805	0.7747	0.2806	0.0638
0.6577	66.0	3300	0.7311	0.805	0.3373	1.0157	0.805	0.7757	0.2880	0.0629
0.6577	67.0	3350	0.7310	0.805	0.3344	1.1878	0.805	0.7766	0.2499	0.0609
0.6577	68.0	3400	0.7326	0.805	0.3391	1.0847	0.805	0.7729	0.2824	0.0636
0.6577	69.0	3450	0.7302	0.805	0.3376	1.1932	0.805	0.7778	0.2789	0.0617
0.6528	70.0	3500	0.7305	0.81	0.3359	0.9988	0.81	0.7787	0.2769	0.0622
0.6528	71.0	3550	0.7300	0.81	0.3328	1.0833	0.81	0.7776	0.2914	0.0594
0.6528	72.0	3600	0.7300	0.81	0.3343	1.1426	0.81	0.7776	0.2843	0.0594
0.6528	73.0	3650	0.7285	0.805	0.3341	1.1237	0.805	0.7701	0.2723	0.0614
0.6528	74.0	3700	0.7303	0.81	0.3368	1.1928	0.81	0.7768	0.2926	0.0612
0.6528	75.0	3750	0.7290	0.805	0.3318	1.0669	0.805	0.7709	0.2810	0.0603
0.6528	76.0	3800	0.7316	0.8	0.3382	1.1392	0.8000	0.7687	0.2505	0.0636
0.6528	77.0	3850	0.7284	0.8	0.3337	1.1338	0.8000	0.7720	0.2677	0.0610
0.6528	78.0	3900	0.7303	0.805	0.3373	1.1969	0.805	0.7729	0.2745	0.0618
0.6528	79.0	3950	0.7297	0.805	0.3369	1.1970	0.805	0.7743	0.2731	0.0606
0.6489	80.0	4000	0.7296	0.795	0.3362	1.1328	0.795	0.7656	0.2620	0.0627
0.6489	81.0	4050	0.7295	0.805	0.3363	1.1358	0.805	0.7726	0.2540	0.0608
0.6489	82.0	4100	0.7290	0.795	0.3341	1.1389	0.795	0.7668	0.2661	0.0630
0.6489	83.0	4150	0.7289	0.8	0.3364	1.0597	0.8000	0.7678	0.2838	0.0615
0.6489	84.0	4200	0.7291	0.805	0.3351	1.1277	0.805	0.7743	0.2621	0.0608
0.6489	85.0	4250	0.7297	0.795	0.3353	1.1953	0.795	0.7668	0.2666	0.0622
0.6489	86.0	4300	0.7286	0.805	0.3339	1.1278	0.805	0.7735	0.2668	0.0608
0.6489	87.0	4350	0.7298	0.8	0.3361	1.1423	0.8000	0.7677	0.2613	0.0614
0.6489	88.0	4400	0.7296	0.805	0.3346	1.1927	0.805	0.7743	0.2789	0.0612
0.6489	89.0	4450	0.7299	0.8	0.3359	1.1950	0.8000	0.7686	0.2500	0.0613
0.6462	90.0	4500	0.7297	0.805	0.3354	1.1934	0.805	0.7743	0.2939	0.0613
0.6462	91.0	4550	0.7294	0.8	0.3353	1.1313	0.8000	0.7685	0.2808	0.0610
0.6462	92.0	4600	0.7297	0.805	0.3356	1.1349	0.805	0.7765	0.2668	0.0614
0.6462	93.0	4650	0.7298	0.8	0.3354	1.1954	0.8000	0.7685	0.2700	0.0613
0.6462	94.0	4700	0.7301	0.8	0.3362	1.1951	0.8000	0.7677	0.2722	0.0616
0.6462	95.0	4750	0.7299	0.805	0.3360	1.1957	0.805	0.7743	0.2619	0.0614
0.6462	96.0	4800	0.7299	0.805	0.3357	1.1946	0.805	0.7743	0.2892	0.0611
0.6462	97.0	4850	0.7297	0.8	0.3355	1.1954	0.8000	0.7686	0.2703	0.0613
0.6462	98.0	4900	0.7298	0.8	0.3359	1.1952	0.8000	0.7677	0.2892	0.0615
0.6462	99.0	4950	0.7298	0.8	0.3357	1.1951	0.8000	0.7677	0.2720	0.0614
0.645	100.0	5000	0.7298	0.8	0.3356	1.1950	0.8000	0.7677	0.2868	0.0614

Framework versions

Transformers 4.36.0.dev0
Pytorch 2.2.0.dev20231112+cu118
Datasets 2.14.5
Tokenizers 0.14.1

jordyvl/dit-base_tobacco-tiny_tobacco3482_simkd

dit-base_tobacco-tiny_tobacco3482_simkd

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions