81-tiny_tobacco3482_kd_NKD_t1.0_g1.5

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: 4.0317
• Accuracy: 0.86
• Brier Loss: 0.2316
• Nll: 0.9681
• F1 Micro: 0.8600
• F1 Macro: 0.8444
• Ece: 0.1162
• Aurc: 0.0505

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: 100

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Brier Loss	Nll	F1 Micro	F1 Macro	Ece	Aurc
No log	1.0	7	6.0999	0.095	1.0191	8.7249	0.095	0.0767	0.3174	0.8988
No log	2.0	14	5.0037	0.145	0.9023	8.1615	0.145	0.1212	0.2431	0.8041
No log	3.0	21	4.5740	0.265	0.8363	5.9714	0.265	0.2003	0.2732	0.5711
No log	4.0	28	4.3460	0.435	0.7292	3.9720	0.435	0.3780	0.3029	0.3535
No log	5.0	35	4.1953	0.565	0.6344	2.7772	0.565	0.4465	0.3056	0.2433
No log	6.0	42	4.0769	0.665	0.5852	2.3975	0.665	0.5345	0.3469	0.1662
No log	7.0	49	3.9670	0.685	0.5103	2.1879	0.685	0.5645	0.3335	0.1339
No log	8.0	56	3.9085	0.74	0.4625	1.7988	0.74	0.6444	0.3276	0.1028
No log	9.0	63	3.8671	0.765	0.4073	1.5650	0.765	0.6567	0.2800	0.0878
No log	10.0	70	3.8008	0.77	0.3617	1.4468	0.7700	0.6730	0.2443	0.0692
No log	11.0	77	3.8924	0.76	0.3685	1.4401	0.76	0.7045	0.2129	0.0859
No log	12.0	84	3.8523	0.78	0.3275	1.3333	0.78	0.7165	0.2191	0.0691
No log	13.0	91	3.7745	0.78	0.3229	1.4554	0.78	0.6820	0.1980	0.0653
No log	14.0	98	3.8155	0.805	0.3070	1.1638	0.805	0.7564	0.1993	0.0621
No log	15.0	105	3.8060	0.815	0.3038	1.2009	0.815	0.7624	0.2051	0.0607
No log	16.0	112	3.8269	0.81	0.3044	1.2937	0.81	0.7751	0.1978	0.0643
No log	17.0	119	3.8191	0.83	0.2841	0.9571	0.83	0.7978	0.1773	0.0593
No log	18.0	126	3.8986	0.81	0.3109	1.1933	0.81	0.7858	0.1728	0.0692
No log	19.0	133	3.8134	0.845	0.2694	0.9314	0.845	0.8175	0.1791	0.0549
No log	20.0	140	3.8148	0.825	0.2768	0.8896	0.825	0.8108	0.1495	0.0619
No log	21.0	147	3.7769	0.83	0.2639	0.9579	0.83	0.8064	0.1521	0.0521
No log	22.0	154	3.7941	0.84	0.2596	0.9457	0.8400	0.8176	0.1644	0.0534
No log	23.0	161	3.9296	0.79	0.2938	1.0757	0.79	0.7732	0.1612	0.0706
No log	24.0	168	3.7899	0.815	0.2734	1.0937	0.815	0.7910	0.1570	0.0531
No log	25.0	175	3.8261	0.855	0.2554	0.9640	0.855	0.8370	0.1378	0.0559
No log	26.0	182	3.8596	0.805	0.2626	1.0429	0.805	0.7814	0.1548	0.0597
No log	27.0	189	3.7974	0.825	0.2637	1.0256	0.825	0.8107	0.1251	0.0522
No log	28.0	196	3.8304	0.83	0.2555	1.1640	0.83	0.8081	0.1337	0.0561
No log	29.0	203	3.8346	0.835	0.2631	0.8774	0.835	0.8110	0.1473	0.0564
No log	30.0	210	3.8251	0.825	0.2549	0.9254	0.825	0.7979	0.1298	0.0541
No log	31.0	217	3.8759	0.825	0.2671	0.9357	0.825	0.8014	0.1672	0.0609
No log	32.0	224	3.8466	0.835	0.2567	1.0822	0.835	0.8168	0.1454	0.0553
No log	33.0	231	3.8600	0.835	0.2578	1.0196	0.835	0.8103	0.1363	0.0577
No log	34.0	238	3.8364	0.84	0.2615	0.9060	0.8400	0.8153	0.1556	0.0550
No log	35.0	245	3.8615	0.84	0.2741	0.9777	0.8400	0.8175	0.1553	0.0609
No log	36.0	252	3.8354	0.815	0.2672	1.0578	0.815	0.7829	0.1412	0.0571
No log	37.0	259	3.8214	0.825	0.2586	1.1244	0.825	0.8005	0.1567	0.0555
No log	38.0	266	3.8379	0.84	0.2557	0.9827	0.8400	0.8159	0.1623	0.0569
No log	39.0	273	3.8269	0.82	0.2590	1.0025	0.82	0.7983	0.1300	0.0552
No log	40.0	280	3.8326	0.835	0.2576	0.9914	0.835	0.8145	0.1339	0.0549
No log	41.0	287	3.8171	0.845	0.2446	0.9369	0.845	0.8234	0.1345	0.0555
No log	42.0	294	3.8197	0.835	0.2421	0.9189	0.835	0.8127	0.1263	0.0551
No log	43.0	301	3.8182	0.835	0.2421	1.0651	0.835	0.8181	0.1348	0.0528
No log	44.0	308	3.8461	0.85	0.2482	0.9057	0.85	0.8353	0.1580	0.0530
No log	45.0	315	3.8203	0.845	0.2405	0.8504	0.845	0.8241	0.1461	0.0512
No log	46.0	322	3.8468	0.845	0.2431	0.7909	0.845	0.8252	0.1273	0.0534
No log	47.0	329	3.8798	0.84	0.2486	0.9810	0.8400	0.8219	0.1412	0.0565
No log	48.0	336	3.8650	0.855	0.2447	0.9117	0.855	0.8372	0.1465	0.0525
No log	49.0	343	3.8425	0.845	0.2414	1.0040	0.845	0.8231	0.1251	0.0536
No log	50.0	350	3.8350	0.86	0.2351	0.8541	0.8600	0.8454	0.1403	0.0511
No log	51.0	357	3.8572	0.84	0.2427	0.9366	0.8400	0.8197	0.1249	0.0527
No log	52.0	364	3.8461	0.85	0.2374	0.8631	0.85	0.8352	0.1269	0.0506
No log	53.0	371	3.8692	0.835	0.2431	0.9839	0.835	0.8110	0.1356	0.0546
No log	54.0	378	3.8806	0.855	0.2403	1.0493	0.855	0.8343	0.1287	0.0530
No log	55.0	385	3.8875	0.86	0.2389	0.9404	0.8600	0.8435	0.1416	0.0532
No log	56.0	392	3.8893	0.84	0.2407	0.9410	0.8400	0.8189	0.1256	0.0539
No log	57.0	399	3.8923	0.84	0.2408	0.9917	0.8400	0.8176	0.1274	0.0540
No log	58.0	406	3.8865	0.86	0.2323	0.9745	0.8600	0.8444	0.1372	0.0517
No log	59.0	413	3.9037	0.845	0.2363	0.9451	0.845	0.8230	0.1422	0.0533
No log	60.0	420	3.9040	0.845	0.2396	1.0699	0.845	0.8233	0.1272	0.0522
No log	61.0	427	3.9136	0.85	0.2378	1.0664	0.85	0.8289	0.1283	0.0527
No log	62.0	434	3.8877	0.85	0.2355	0.8483	0.85	0.8315	0.1445	0.0519
No log	63.0	441	3.8991	0.855	0.2342	0.9246	0.855	0.8374	0.1124	0.0509
No log	64.0	448	3.9207	0.85	0.2383	1.0477	0.85	0.8330	0.1235	0.0514
No log	65.0	455	3.9000	0.855	0.2362	1.0504	0.855	0.8384	0.1375	0.0529
No log	66.0	462	3.9542	0.84	0.2463	0.9763	0.8400	0.8255	0.1405	0.0540
No log	67.0	469	3.9153	0.855	0.2374	0.9954	0.855	0.8376	0.1353	0.0523
No log	68.0	476	3.9264	0.845	0.2410	0.9917	0.845	0.8295	0.1081	0.0515
No log	69.0	483	3.8989	0.86	0.2272	0.9322	0.8600	0.8438	0.1260	0.0492
No log	70.0	490	3.9224	0.86	0.2329	0.9317	0.8600	0.8443	0.1091	0.0515
No log	71.0	497	3.9313	0.85	0.2360	1.0424	0.85	0.8310	0.1259	0.0511
3.5118	72.0	504	3.9407	0.85	0.2343	0.9333	0.85	0.8331	0.1156	0.0529
3.5118	73.0	511	3.9407	0.865	0.2318	0.9791	0.865	0.8523	0.1245	0.0497
3.5118	74.0	518	3.9461	0.855	0.2347	1.0488	0.855	0.8385	0.1298	0.0508
3.5118	75.0	525	3.9560	0.86	0.2319	0.9924	0.8600	0.8444	0.1410	0.0504
3.5118	76.0	532	3.9608	0.855	0.2317	0.9253	0.855	0.8390	0.1380	0.0517
3.5118	77.0	539	3.9638	0.865	0.2319	0.9210	0.865	0.8528	0.1202	0.0504
3.5118	78.0	546	3.9718	0.86	0.2323	0.9413	0.8600	0.8444	0.1255	0.0505
3.5118	79.0	553	3.9778	0.86	0.2324	0.9916	0.8600	0.8444	0.1270	0.0506
3.5118	80.0	560	3.9813	0.855	0.2323	0.9919	0.855	0.8390	0.1246	0.0509
3.5118	81.0	567	3.9876	0.86	0.2319	0.9330	0.8600	0.8444	0.1318	0.0506
3.5118	82.0	574	3.9939	0.855	0.2324	0.9328	0.855	0.8390	0.1280	0.0510
3.5118	83.0	581	3.9971	0.86	0.2319	0.9321	0.8600	0.8444	0.1303	0.0503
3.5118	84.0	588	4.0003	0.855	0.2316	0.9348	0.855	0.8390	0.1284	0.0508
3.5118	85.0	595	4.0054	0.86	0.2319	0.9909	0.8600	0.8444	0.1348	0.0503
3.5118	86.0	602	4.0086	0.86	0.2315	0.9338	0.8600	0.8444	0.1340	0.0504
3.5118	87.0	609	4.0125	0.86	0.2318	0.9522	0.8600	0.8444	0.1348	0.0504
3.5118	88.0	616	4.0148	0.86	0.2316	0.9396	0.8600	0.8444	0.1323	0.0504
3.5118	89.0	623	4.0185	0.86	0.2318	0.9378	0.8600	0.8444	0.1326	0.0505
3.5118	90.0	630	4.0197	0.86	0.2316	0.9412	0.8600	0.8444	0.1253	0.0506
3.5118	91.0	637	4.0231	0.86	0.2318	0.9395	0.8600	0.8444	0.1165	0.0506
3.5118	92.0	644	4.0249	0.86	0.2316	0.9921	0.8600	0.8444	0.1159	0.0504
3.5118	93.0	651	4.0266	0.86	0.2316	0.9441	0.8600	0.8444	0.1161	0.0505
3.5118	94.0	658	4.0275	0.86	0.2317	0.9934	0.8600	0.8444	0.1159	0.0504
3.5118	95.0	665	4.0289	0.86	0.2315	0.9429	0.8600	0.8444	0.1160	0.0505
3.5118	96.0	672	4.0301	0.86	0.2316	0.9932	0.8600	0.8444	0.1163	0.0505
3.5118	97.0	679	4.0304	0.86	0.2315	0.9936	0.8600	0.8444	0.1163	0.0505
3.5118	98.0	686	4.0313	0.86	0.2316	0.9935	0.8600	0.8444	0.1163	0.0504
3.5118	99.0	693	4.0317	0.86	0.2316	0.9601	0.8600	0.8444	0.1162	0.0505
3.5118	100.0	700	4.0317	0.86	0.2316	0.9681	0.8600	0.8444	0.1162	0.0505

Framework versions

• Transformers 4.26.1
• Pytorch 1.13.1.post200
• Datasets 2.9.0
• Tokenizers 0.13.2