dit-base_tobacco-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: 3.1418
• Accuracy: 0.84
• Brier Loss: 0.2718
• Nll: 0.9778
• F1 Micro: 0.8400
• F1 Macro: 0.8296
• Ece: 0.1553
• Aurc: 0.0479

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	5.6749	0.2	0.9075	8.1551	0.2000	0.1380	0.2949	0.8075
No log	2.0	14	5.1602	0.15	0.8781	7.4212	0.15	0.1415	0.2402	0.7606
No log	3.0	21	4.4947	0.46	0.7066	2.8419	0.46	0.4202	0.3230	0.3244
No log	4.0	28	3.9789	0.555	0.5827	1.6986	0.555	0.5525	0.2796	0.2218
No log	5.0	35	3.6991	0.65	0.4828	1.7197	0.65	0.6491	0.2315	0.1491
No log	6.0	42	3.6495	0.68	0.4691	1.8258	0.68	0.6586	0.2555	0.1358
No log	7.0	49	3.3912	0.75	0.3899	1.8385	0.75	0.7276	0.2237	0.0920
No log	8.0	56	3.3055	0.71	0.3792	1.5754	0.7100	0.6922	0.2130	0.1023
No log	9.0	63	3.3535	0.72	0.3836	1.7076	0.72	0.7195	0.2015	0.0978
No log	10.0	70	3.1877	0.785	0.3190	1.5736	0.785	0.7582	0.1905	0.0693
No log	11.0	77	3.5578	0.72	0.3812	2.2613	0.72	0.7241	0.1684	0.0846
No log	12.0	84	3.3589	0.775	0.3389	1.3228	0.775	0.7540	0.1665	0.0764
No log	13.0	91	3.1097	0.805	0.2856	1.4183	0.805	0.7929	0.1603	0.0484
No log	14.0	98	3.2661	0.815	0.3146	1.9097	0.815	0.8066	0.1753	0.0636
No log	15.0	105	3.3637	0.755	0.3361	1.5166	0.755	0.7492	0.1804	0.0720
No log	16.0	112	3.1495	0.8	0.2994	1.4586	0.8000	0.7926	0.1714	0.0604
No log	17.0	119	3.1573	0.8	0.2941	1.6755	0.8000	0.7899	0.1545	0.0577
No log	18.0	126	3.4445	0.77	0.3416	1.4075	0.7700	0.7503	0.1620	0.0807
No log	19.0	133	3.1292	0.805	0.2816	1.3835	0.805	0.7815	0.1768	0.0526
No log	20.0	140	3.4253	0.75	0.3459	2.0430	0.75	0.7591	0.1697	0.0706
No log	21.0	147	3.1645	0.81	0.3000	1.7363	0.81	0.8113	0.1711	0.0614
No log	22.0	154	3.0823	0.815	0.2791	1.5997	0.815	0.8020	0.1417	0.0556
No log	23.0	161	2.9898	0.83	0.2521	1.4274	0.83	0.8189	0.1589	0.0434
No log	24.0	168	3.0915	0.83	0.2770	1.3516	0.83	0.8173	0.1495	0.0538
No log	25.0	175	3.0405	0.825	0.2621	1.5191	0.825	0.8048	0.1329	0.0494
No log	26.0	182	3.0621	0.815	0.2735	1.0698	0.815	0.7955	0.1617	0.0522
No log	27.0	189	3.0228	0.835	0.2650	1.4235	0.835	0.8315	0.1565	0.0502
No log	28.0	196	3.0677	0.82	0.2778	1.5299	0.82	0.8165	0.1660	0.0557
No log	29.0	203	3.0272	0.825	0.2699	1.4726	0.825	0.8204	0.1643	0.0491
No log	30.0	210	3.1090	0.815	0.2892	1.3258	0.815	0.8026	0.1585	0.0536
No log	31.0	217	3.1069	0.81	0.2866	1.5638	0.81	0.8050	0.1473	0.0557
No log	32.0	224	3.0374	0.815	0.2765	1.2895	0.815	0.8045	0.1476	0.0527
No log	33.0	231	3.0503	0.815	0.2750	1.3113	0.815	0.7975	0.1531	0.0517
No log	34.0	238	2.9852	0.82	0.2613	1.4575	0.82	0.8110	0.1600	0.0448
No log	35.0	245	3.0437	0.83	0.2724	1.3491	0.83	0.8205	0.1622	0.0571
No log	36.0	252	3.0098	0.82	0.2717	1.2671	0.82	0.8055	0.1567	0.0519
No log	37.0	259	3.0025	0.845	0.2599	1.2628	0.845	0.8255	0.1342	0.0481
No log	38.0	266	3.1854	0.805	0.3015	1.2550	0.805	0.7956	0.1560	0.0601
No log	39.0	273	3.0704	0.82	0.2793	1.2393	0.82	0.8057	0.1566	0.0557
No log	40.0	280	3.0739	0.825	0.2842	1.2701	0.825	0.8169	0.1371	0.0513
No log	41.0	287	3.0465	0.835	0.2747	1.2598	0.835	0.8302	0.1449	0.0538
No log	42.0	294	3.0691	0.825	0.2773	1.1796	0.825	0.8137	0.1372	0.0511
No log	43.0	301	3.0734	0.84	0.2732	1.1765	0.8400	0.8282	0.1564	0.0565
No log	44.0	308	3.0262	0.845	0.2622	1.2152	0.845	0.8306	0.1457	0.0541
No log	45.0	315	3.0610	0.835	0.2727	1.2249	0.835	0.8261	0.1606	0.0544
No log	46.0	322	3.0358	0.84	0.2767	1.1020	0.8400	0.8323	0.1416	0.0527
No log	47.0	329	2.9893	0.835	0.2650	1.1536	0.835	0.8252	0.1386	0.0493
No log	48.0	336	3.0498	0.84	0.2726	1.1253	0.8400	0.8320	0.1302	0.0535
No log	49.0	343	2.9816	0.845	0.2585	1.2068	0.845	0.8355	0.1455	0.0451
No log	50.0	350	3.0431	0.835	0.2686	1.0596	0.835	0.8238	0.1542	0.0540
No log	51.0	357	3.0200	0.835	0.2639	1.1806	0.835	0.8290	0.1434	0.0501
No log	52.0	364	3.0217	0.845	0.2664	1.0846	0.845	0.8324	0.1671	0.0503
No log	53.0	371	3.0255	0.84	0.2649	1.1803	0.8400	0.8318	0.1350	0.0488
No log	54.0	378	3.0069	0.835	0.2616	1.2057	0.835	0.8190	0.1284	0.0496
No log	55.0	385	3.0609	0.815	0.2746	1.0378	0.815	0.7970	0.1422	0.0490
No log	56.0	392	3.0111	0.84	0.2622	1.1806	0.8400	0.8341	0.1428	0.0513
No log	57.0	399	3.0050	0.84	0.2643	1.1898	0.8400	0.8299	0.1452	0.0494
No log	58.0	406	3.0426	0.84	0.2662	1.0337	0.8400	0.8307	0.1397	0.0514
No log	59.0	413	3.0427	0.835	0.2682	1.0309	0.835	0.8247	0.1453	0.0491
No log	60.0	420	3.0449	0.83	0.2744	1.0039	0.83	0.8141	0.1436	0.0484
No log	61.0	427	3.0524	0.83	0.2729	1.1480	0.83	0.8162	0.1454	0.0477
No log	62.0	434	3.0290	0.835	0.2610	1.1757	0.835	0.8264	0.1476	0.0506
No log	63.0	441	3.0574	0.83	0.2712	1.0242	0.83	0.8161	0.1464	0.0485
No log	64.0	448	3.0436	0.835	0.2684	1.1326	0.835	0.8267	0.1417	0.0470
No log	65.0	455	3.0170	0.84	0.2610	1.1095	0.8400	0.8289	0.1520	0.0492
No log	66.0	462	3.0176	0.835	0.2623	1.1140	0.835	0.8225	0.1262	0.0459
No log	67.0	469	3.0712	0.84	0.2735	1.0884	0.8400	0.8296	0.1421	0.0516
No log	68.0	476	3.0258	0.84	0.2670	1.1388	0.8400	0.8279	0.1478	0.0461
No log	69.0	483	3.0838	0.835	0.2707	1.0937	0.835	0.8232	0.1425	0.0477
No log	70.0	490	3.1076	0.82	0.2819	1.0030	0.82	0.7998	0.1507	0.0480
No log	71.0	497	3.0696	0.84	0.2725	1.0175	0.8400	0.8349	0.1567	0.0501
2.6485	72.0	504	3.0535	0.84	0.2676	1.0079	0.8400	0.8253	0.1351	0.0477
2.6485	73.0	511	3.0326	0.83	0.2667	0.9792	0.83	0.8093	0.1334	0.0464
2.6485	74.0	518	3.0271	0.835	0.2616	1.0865	0.835	0.8193	0.1223	0.0454
2.6485	75.0	525	3.0894	0.83	0.2732	0.9764	0.83	0.8123	0.1446	0.0489
2.6485	76.0	532	3.0905	0.835	0.2730	1.0736	0.835	0.8232	0.1578	0.0485
2.6485	77.0	539	3.0507	0.84	0.2646	1.0716	0.8400	0.8279	0.1424	0.0469
2.6485	78.0	546	3.0981	0.845	0.2712	0.9916	0.845	0.8324	0.1452	0.0508
2.6485	79.0	553	3.0820	0.84	0.2728	0.9791	0.8400	0.8296	0.1403	0.0473
2.6485	80.0	560	3.0978	0.84	0.2733	0.9864	0.8400	0.8296	0.1480	0.0485
2.6485	81.0	567	3.0936	0.84	0.2716	0.9955	0.8400	0.8296	0.1483	0.0474
2.6485	82.0	574	3.0937	0.845	0.2685	0.9875	0.845	0.8324	0.1459	0.0486
2.6485	83.0	581	3.0940	0.84	0.2719	0.9863	0.8400	0.8296	0.1481	0.0470
2.6485	84.0	588	3.0745	0.84	0.2656	1.0795	0.8400	0.8323	0.1460	0.0476
2.6485	85.0	595	3.1089	0.845	0.2681	1.0050	0.845	0.8324	0.1568	0.0492
2.6485	86.0	602	3.0880	0.84	0.2695	1.0607	0.8400	0.8296	0.1409	0.0474
2.6485	87.0	609	3.0848	0.84	0.2666	0.9996	0.8400	0.8296	0.1425	0.0470
2.6485	88.0	616	3.1144	0.84	0.2682	0.9937	0.8400	0.8296	0.1380	0.0482
2.6485	89.0	623	3.1316	0.84	0.2711	0.9884	0.8400	0.8296	0.1484	0.0490
2.6485	90.0	630	3.1312	0.84	0.2726	0.9732	0.8400	0.8296	0.1525	0.0488
2.6485	91.0	637	3.1312	0.84	0.2723	0.9794	0.8400	0.8296	0.1475	0.0481
2.6485	92.0	644	3.1426	0.84	0.2731	0.9728	0.8400	0.8296	0.1478	0.0491
2.6485	93.0	651	3.1351	0.84	0.2709	0.9741	0.8400	0.8296	0.1438	0.0483
2.6485	94.0	658	3.1390	0.84	0.2716	0.9764	0.8400	0.8296	0.1576	0.0483
2.6485	95.0	665	3.1366	0.84	0.2711	0.9795	0.8400	0.8296	0.1480	0.0484
2.6485	96.0	672	3.1337	0.84	0.2710	0.9828	0.8400	0.8296	0.1475	0.0478
2.6485	97.0	679	3.1431	0.84	0.2723	0.9767	0.8400	0.8296	0.1587	0.0480
2.6485	98.0	686	3.1388	0.84	0.2713	0.9808	0.8400	0.8296	0.1476	0.0480
2.6485	99.0	693	3.1420	0.84	0.2718	0.9778	0.8400	0.8296	0.1560	0.0480
2.6485	100.0	700	3.1418	0.84	0.2718	0.9778	0.8400	0.8296	0.1553	0.0479

Framework versions

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