dit-base_tobacco-small_tobacco3482_kd_CEKD_t2.5_a0.5

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: 0.6146
• Accuracy: 0.8
• Brier Loss: 0.2784
• Nll: 1.4268
• F1 Micro: 0.8000
• F1 Macro: 0.7846
• Ece: 0.1626
• Aurc: 0.0474

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	4.1581	0.18	0.8974	4.2254	0.18	0.1559	0.2651	0.8061
No log	2.0	14	3.2929	0.355	0.7710	4.0541	0.3550	0.2167	0.2742	0.4326
No log	3.0	21	2.2155	0.55	0.5837	2.0462	0.55	0.4296	0.2323	0.2481
No log	4.0	28	1.5197	0.7	0.4370	1.7716	0.7	0.6411	0.2342	0.1327
No log	5.0	35	1.2831	0.715	0.4289	1.7142	0.715	0.6859	0.2047	0.1211
No log	6.0	42	1.2204	0.72	0.3989	1.6102	0.72	0.6999	0.1961	0.1066
No log	7.0	49	0.9767	0.755	0.3317	1.5919	0.755	0.7148	0.1724	0.0775
No log	8.0	56	0.8875	0.785	0.3049	1.4209	0.785	0.7633	0.1478	0.0716
No log	9.0	63	0.9311	0.79	0.3185	1.5420	0.79	0.7474	0.1645	0.0741
No log	10.0	70	0.8116	0.835	0.2672	1.5127	0.835	0.8232	0.1463	0.0563
No log	11.0	77	0.8315	0.805	0.3054	1.6275	0.805	0.7897	0.1695	0.0618
No log	12.0	84	0.7678	0.815	0.2917	1.5009	0.815	0.8012	0.1469	0.0542
No log	13.0	91	0.7249	0.81	0.2816	1.4685	0.81	0.7880	0.1437	0.0576
No log	14.0	98	0.8116	0.815	0.2894	1.5975	0.815	0.7941	0.1481	0.0604
No log	15.0	105	0.7985	0.81	0.3098	1.4721	0.81	0.7819	0.1646	0.0662
No log	16.0	112	0.6839	0.815	0.2781	1.4357	0.815	0.7992	0.1589	0.0529
No log	17.0	119	0.6590	0.82	0.2670	1.4487	0.82	0.8061	0.1336	0.0461
No log	18.0	126	0.7253	0.81	0.2938	1.5163	0.81	0.7951	0.1617	0.0558
No log	19.0	133	0.6935	0.795	0.2949	1.4516	0.795	0.7758	0.1736	0.0531
No log	20.0	140	0.6991	0.795	0.2875	1.3932	0.795	0.7735	0.1584	0.0519
No log	21.0	147	0.7059	0.815	0.2966	1.5011	0.815	0.7927	0.1579	0.0565
No log	22.0	154	0.6754	0.79	0.2896	1.4549	0.79	0.7742	0.1534	0.0531
No log	23.0	161	0.6981	0.785	0.2989	1.4261	0.785	0.7705	0.1490	0.0530
No log	24.0	168	0.6503	0.805	0.2842	1.4998	0.805	0.7885	0.1415	0.0512
No log	25.0	175	0.6680	0.79	0.2891	1.4228	0.79	0.7742	0.1504	0.0519
No log	26.0	182	0.6835	0.81	0.2948	1.4400	0.81	0.7944	0.1545	0.0516
No log	27.0	189	0.6495	0.81	0.2846	1.4433	0.81	0.7868	0.1552	0.0503
No log	28.0	196	0.6450	0.81	0.2851	1.4037	0.81	0.7913	0.1476	0.0498
No log	29.0	203	0.6634	0.815	0.2861	1.4186	0.815	0.7966	0.1397	0.0521
No log	30.0	210	0.6212	0.805	0.2739	1.4265	0.805	0.7902	0.1444	0.0482
No log	31.0	217	0.6271	0.815	0.2800	1.4392	0.815	0.7986	0.1370	0.0494
No log	32.0	224	0.6256	0.8	0.2786	1.3677	0.8000	0.7811	0.1454	0.0496
No log	33.0	231	0.6219	0.805	0.2779	1.4276	0.805	0.7857	0.1580	0.0465
No log	34.0	238	0.6203	0.81	0.2779	1.4392	0.81	0.7914	0.1275	0.0470
No log	35.0	245	0.6193	0.81	0.2793	1.4258	0.81	0.7934	0.1438	0.0483
No log	36.0	252	0.6261	0.83	0.2743	1.4227	0.83	0.8098	0.1482	0.0501
No log	37.0	259	0.6190	0.815	0.2776	1.4301	0.815	0.7977	0.1446	0.0484
No log	38.0	266	0.6210	0.805	0.2867	1.4958	0.805	0.7878	0.1477	0.0496
No log	39.0	273	0.5974	0.805	0.2771	1.5068	0.805	0.7901	0.1381	0.0476
No log	40.0	280	0.6224	0.8	0.2869	1.4325	0.8000	0.7869	0.1443	0.0472
No log	41.0	287	0.6178	0.805	0.2796	1.4316	0.805	0.7912	0.1454	0.0471
No log	42.0	294	0.6194	0.825	0.2765	1.5001	0.825	0.8059	0.1401	0.0474
No log	43.0	301	0.6224	0.805	0.2769	1.4268	0.805	0.7888	0.1398	0.0493
No log	44.0	308	0.6265	0.8	0.2819	1.4401	0.8000	0.7846	0.1422	0.0481
No log	45.0	315	0.6275	0.8	0.2819	1.4206	0.8000	0.7847	0.1465	0.0487
No log	46.0	322	0.6173	0.805	0.2806	1.3618	0.805	0.7870	0.1383	0.0478
No log	47.0	329	0.6177	0.81	0.2804	1.4988	0.81	0.7906	0.1468	0.0488
No log	48.0	336	0.6175	0.81	0.2788	1.4356	0.81	0.7917	0.1460	0.0476
No log	49.0	343	0.6209	0.81	0.2775	1.4290	0.81	0.7925	0.1603	0.0478
No log	50.0	350	0.6244	0.815	0.2780	1.3662	0.815	0.7974	0.1322	0.0480
No log	51.0	357	0.6176	0.81	0.2777	1.4307	0.81	0.7941	0.1258	0.0478
No log	52.0	364	0.6150	0.805	0.2774	1.4310	0.805	0.7896	0.1369	0.0477
No log	53.0	371	0.6164	0.81	0.2772	1.4298	0.81	0.7941	0.1391	0.0479
No log	54.0	378	0.6137	0.81	0.2766	1.4291	0.81	0.7928	0.1358	0.0474
No log	55.0	385	0.6163	0.81	0.2776	1.4298	0.81	0.7928	0.1278	0.0475
No log	56.0	392	0.6148	0.81	0.2776	1.4286	0.81	0.7928	0.1480	0.0471
No log	57.0	399	0.6154	0.81	0.2773	1.4290	0.81	0.7928	0.1485	0.0474
No log	58.0	406	0.6143	0.8	0.2781	1.4281	0.8000	0.7852	0.1405	0.0473
No log	59.0	413	0.6158	0.805	0.2785	1.4295	0.805	0.7899	0.1455	0.0473
No log	60.0	420	0.6146	0.805	0.2774	1.4310	0.805	0.7899	0.1346	0.0472
No log	61.0	427	0.6154	0.805	0.2780	1.4292	0.805	0.7899	0.1451	0.0472
No log	62.0	434	0.6148	0.805	0.2780	1.4304	0.805	0.7905	0.1543	0.0473
No log	63.0	441	0.6150	0.8	0.2783	1.4284	0.8000	0.7846	0.1502	0.0473
No log	64.0	448	0.6143	0.805	0.2780	1.4294	0.805	0.7899	0.1453	0.0470
No log	65.0	455	0.6152	0.805	0.2782	1.4298	0.805	0.7899	0.1373	0.0469
No log	66.0	462	0.6148	0.8	0.2781	1.4287	0.8000	0.7852	0.1492	0.0475
No log	67.0	469	0.6134	0.805	0.2776	1.4286	0.805	0.7899	0.1526	0.0470
No log	68.0	476	0.6150	0.8	0.2785	1.4270	0.8000	0.7846	0.1497	0.0474
No log	69.0	483	0.6145	0.8	0.2783	1.4281	0.8000	0.7846	0.1483	0.0471
No log	70.0	490	0.6145	0.805	0.2778	1.4292	0.805	0.7899	0.1472	0.0471
No log	71.0	497	0.6143	0.805	0.2779	1.4284	0.805	0.7899	0.1529	0.0470
0.2616	72.0	504	0.6148	0.805	0.2780	1.4276	0.805	0.7899	0.1414	0.0471
0.2616	73.0	511	0.6147	0.8	0.2781	1.4285	0.8000	0.7852	0.1400	0.0473
0.2616	74.0	518	0.6147	0.8	0.2783	1.4281	0.8000	0.7846	0.1501	0.0473
0.2616	75.0	525	0.6150	0.8	0.2784	1.4269	0.8000	0.7846	0.1417	0.0473
0.2616	76.0	532	0.6143	0.805	0.2782	1.4273	0.805	0.7899	0.1524	0.0470
0.2616	77.0	539	0.6147	0.805	0.2782	1.4277	0.805	0.7899	0.1526	0.0470
0.2616	78.0	546	0.6149	0.8	0.2785	1.4277	0.8000	0.7846	0.1572	0.0474
0.2616	79.0	553	0.6147	0.805	0.2782	1.4276	0.805	0.7899	0.1529	0.0471
0.2616	80.0	560	0.6145	0.805	0.2783	1.4278	0.805	0.7899	0.1527	0.0471
0.2616	81.0	567	0.6147	0.8	0.2783	1.4277	0.8000	0.7846	0.1483	0.0472
0.2616	82.0	574	0.6146	0.8	0.2783	1.4275	0.8000	0.7846	0.1623	0.0473
0.2616	83.0	581	0.6145	0.8	0.2783	1.4274	0.8000	0.7846	0.1571	0.0473
0.2616	84.0	588	0.6146	0.8	0.2782	1.4276	0.8000	0.7846	0.1538	0.0473
0.2616	85.0	595	0.6146	0.805	0.2783	1.4274	0.805	0.7899	0.1493	0.0471
0.2616	86.0	602	0.6147	0.8	0.2784	1.4269	0.8000	0.7846	0.1627	0.0473
0.2616	87.0	609	0.6146	0.8	0.2783	1.4270	0.8000	0.7846	0.1623	0.0472
0.2616	88.0	616	0.6145	0.805	0.2783	1.4272	0.805	0.7899	0.1579	0.0470
0.2616	89.0	623	0.6146	0.8	0.2784	1.4272	0.8000	0.7846	0.1627	0.0474
0.2616	90.0	630	0.6147	0.8	0.2783	1.4270	0.8000	0.7846	0.1536	0.0473
0.2616	91.0	637	0.6147	0.8	0.2784	1.4268	0.8000	0.7846	0.1627	0.0475
0.2616	92.0	644	0.6145	0.805	0.2783	1.4268	0.805	0.7899	0.1582	0.0471
0.2616	93.0	651	0.6145	0.8	0.2784	1.4269	0.8000	0.7846	0.1626	0.0474
0.2616	94.0	658	0.6146	0.8	0.2784	1.4268	0.8000	0.7846	0.1626	0.0473
0.2616	95.0	665	0.6147	0.8	0.2784	1.4268	0.8000	0.7846	0.1626	0.0473
0.2616	96.0	672	0.6146	0.8	0.2784	1.4269	0.8000	0.7846	0.1626	0.0474
0.2616	97.0	679	0.6146	0.8	0.2784	1.4269	0.8000	0.7846	0.1626	0.0474
0.2616	98.0	686	0.6146	0.8	0.2784	1.4269	0.8000	0.7846	0.1626	0.0474
0.2616	99.0	693	0.6146	0.8	0.2784	1.4268	0.8000	0.7846	0.1626	0.0474
0.2616	100.0	700	0.6146	0.8	0.2784	1.4268	0.8000	0.7846	0.1626	0.0474

Framework versions

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