18-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: 4.0957
Accuracy: 0.805
Brier Loss: 0.2927
Nll: 1.1753
F1 Micro: 0.805
F1 Macro: 0.7833
Ece: 0.1572
Aurc: 0.0655

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	4.7898	0.1	1.0292	9.4902	0.1000	0.0772	0.3220	0.9001
No log	2.0	14	3.9970	0.1	0.9420	10.0981	0.1000	0.1071	0.2441	0.8581
No log	3.0	21	3.6641	0.075	0.8956	9.5324	0.075	0.0777	0.1896	0.9137
No log	4.0	28	3.6014	0.18	0.8691	9.6679	0.18	0.0781	0.2345	0.5824
No log	5.0	35	3.5833	0.23	0.8347	9.6569	0.23	0.1572	0.2618	0.5205
No log	6.0	42	3.5576	0.44	0.7860	5.9410	0.44	0.2946	0.3475	0.3232
No log	7.0	49	3.5400	0.575	0.7404	4.2387	0.575	0.4638	0.4007	0.2294
No log	8.0	56	3.5319	0.545	0.7181	4.5958	0.545	0.4482	0.3502	0.2374
No log	9.0	63	3.5405	0.52	0.7002	3.9862	0.52	0.4101	0.3148	0.2506
No log	10.0	70	3.5341	0.61	0.6897	3.2707	0.61	0.5118	0.3775	0.2235
No log	11.0	77	3.5259	0.66	0.6771	2.6882	0.66	0.5201	0.4365	0.1420
No log	12.0	84	3.5215	0.66	0.6463	2.4544	0.66	0.5387	0.3750	0.1664
No log	13.0	91	3.5363	0.58	0.6232	2.3149	0.58	0.5090	0.3285	0.1858
No log	14.0	98	3.5161	0.675	0.6008	2.6144	0.675	0.5411	0.3690	0.1237
No log	15.0	105	3.5073	0.67	0.5845	2.1229	0.67	0.5577	0.3405	0.1350
No log	16.0	112	3.5272	0.67	0.5338	2.4215	0.67	0.5603	0.3154	0.1325
No log	17.0	119	3.5332	0.695	0.5367	2.1675	0.695	0.6056	0.3140	0.1071
No log	18.0	126	3.5659	0.655	0.4841	1.9565	0.655	0.5559	0.2600	0.1365
No log	19.0	133	3.5438	0.69	0.4817	1.8201	0.69	0.5735	0.2574	0.1202
No log	20.0	140	3.5019	0.74	0.4725	1.6346	0.74	0.6486	0.2939	0.0931
No log	21.0	147	3.5236	0.755	0.4407	1.3134	0.755	0.6811	0.2762	0.0820
No log	22.0	154	3.5303	0.755	0.4143	1.2834	0.755	0.6843	0.2434	0.0806
No log	23.0	161	3.5541	0.77	0.4034	1.4417	0.7700	0.6891	0.2382	0.0842
No log	24.0	168	3.5675	0.765	0.3853	1.6692	0.765	0.7072	0.2309	0.0807
No log	25.0	175	3.5411	0.745	0.3914	1.2777	0.745	0.6720	0.2271	0.0784
No log	26.0	182	3.5877	0.75	0.3710	1.4838	0.75	0.6717	0.2082	0.0789
No log	27.0	189	3.6026	0.77	0.3483	1.4211	0.7700	0.7018	0.2089	0.0694
No log	28.0	196	3.6374	0.78	0.3365	1.3205	0.78	0.7181	0.1953	0.0694
No log	29.0	203	3.7319	0.775	0.3538	1.2749	0.775	0.7012	0.2149	0.0814
No log	30.0	210	3.6359	0.805	0.3291	1.3272	0.805	0.7761	0.1991	0.0637
No log	31.0	217	3.7160	0.785	0.3337	1.2632	0.785	0.7445	0.1727	0.0757
No log	32.0	224	3.6810	0.8	0.3234	1.3720	0.8000	0.7636	0.1999	0.0649
No log	33.0	231	3.7139	0.82	0.3221	1.2150	0.82	0.7919	0.2051	0.0677
No log	34.0	238	3.7286	0.795	0.3130	1.0622	0.795	0.7575	0.1919	0.0639
No log	35.0	245	3.7807	0.795	0.3154	1.0146	0.795	0.7672	0.1565	0.0714
No log	36.0	252	3.6802	0.815	0.3131	1.0083	0.815	0.7933	0.2051	0.0626
No log	37.0	259	3.7369	0.81	0.3168	1.0017	0.81	0.7862	0.1792	0.0690
No log	38.0	266	3.7638	0.82	0.2971	1.3357	0.82	0.7977	0.1913	0.0628
No log	39.0	273	3.7415	0.825	0.2954	1.0423	0.825	0.8072	0.1893	0.0599
No log	40.0	280	3.8005	0.785	0.3140	1.0817	0.785	0.7453	0.1694	0.0684
No log	41.0	287	3.7901	0.82	0.3127	1.0853	0.82	0.7993	0.1789	0.0673
No log	42.0	294	3.7811	0.825	0.3019	1.2712	0.825	0.8020	0.1644	0.0644
No log	43.0	301	3.7689	0.81	0.3110	0.8553	0.81	0.7932	0.1785	0.0645
No log	44.0	308	3.7796	0.82	0.2919	1.2589	0.82	0.7972	0.1875	0.0643
No log	45.0	315	3.8005	0.805	0.3036	1.1993	0.805	0.7789	0.1840	0.0660
No log	46.0	322	3.7811	0.82	0.2909	1.0962	0.82	0.8004	0.1735	0.0618
No log	47.0	329	3.8145	0.8	0.3040	1.1968	0.8000	0.7759	0.1795	0.0671
No log	48.0	336	3.7969	0.835	0.2816	1.1019	0.835	0.8118	0.1624	0.0603
No log	49.0	343	3.8020	0.815	0.2855	1.0383	0.815	0.7978	0.1556	0.0639
No log	50.0	350	3.8049	0.815	0.2884	1.1121	0.815	0.7935	0.1608	0.0616
No log	51.0	357	3.8048	0.81	0.2873	1.1173	0.81	0.7898	0.1574	0.0632
No log	52.0	364	3.8581	0.8	0.2923	1.1257	0.8000	0.7767	0.1436	0.0664
No log	53.0	371	3.8565	0.79	0.2984	1.0513	0.79	0.7670	0.1622	0.0668
No log	54.0	378	3.8787	0.805	0.2901	1.0619	0.805	0.7874	0.1335	0.0655
No log	55.0	385	3.8777	0.805	0.2940	1.0378	0.805	0.7883	0.1450	0.0647
No log	56.0	392	3.8743	0.805	0.2906	1.1702	0.805	0.7849	0.1610	0.0634
No log	57.0	399	3.9082	0.795	0.2959	1.0951	0.795	0.7711	0.1761	0.0662
No log	58.0	406	3.8894	0.8	0.2898	1.0979	0.8000	0.7816	0.1774	0.0638
No log	59.0	413	3.9005	0.825	0.2914	1.2358	0.825	0.8088	0.1687	0.0637
No log	60.0	420	3.9115	0.815	0.2863	1.0318	0.815	0.7928	0.1672	0.0640
No log	61.0	427	3.9172	0.805	0.2956	1.1397	0.805	0.7884	0.1646	0.0667
No log	62.0	434	3.8993	0.82	0.2862	1.2349	0.82	0.8001	0.1544	0.0645
No log	63.0	441	3.9334	0.825	0.2896	1.1718	0.825	0.8061	0.1662	0.0646
No log	64.0	448	3.9179	0.815	0.2861	1.1727	0.815	0.7966	0.1592	0.0650
No log	65.0	455	3.9489	0.8	0.2981	1.1681	0.8000	0.7805	0.1522	0.0674
No log	66.0	462	3.9372	0.81	0.2855	1.1041	0.81	0.7870	0.1709	0.0647
No log	67.0	469	3.9651	0.8	0.2935	1.1723	0.8000	0.7816	0.1492	0.0667
No log	68.0	476	3.9600	0.815	0.2903	1.1687	0.815	0.7950	0.1466	0.0650
No log	69.0	483	3.9695	0.82	0.2908	1.1251	0.82	0.8026	0.1532	0.0654
No log	70.0	490	3.9817	0.805	0.2915	1.1879	0.805	0.7861	0.1537	0.0657
No log	71.0	497	3.9838	0.81	0.2899	1.1688	0.81	0.7892	0.1538	0.0648
3.4085	72.0	504	3.9960	0.805	0.2910	1.1702	0.805	0.7904	0.1568	0.0657
3.4085	73.0	511	4.0046	0.8	0.2931	1.1743	0.8000	0.7800	0.1529	0.0658
3.4085	74.0	518	4.0115	0.815	0.2917	1.1718	0.815	0.7968	0.1589	0.0647
3.4085	75.0	525	4.0205	0.805	0.2920	1.1719	0.805	0.7833	0.1575	0.0654
3.4085	76.0	532	4.0272	0.805	0.2919	1.1725	0.805	0.7833	0.1547	0.0659
3.4085	77.0	539	4.0323	0.81	0.2923	1.1720	0.81	0.7892	0.1547	0.0653
3.4085	78.0	546	4.0364	0.81	0.2907	1.1715	0.81	0.7892	0.1607	0.0650
3.4085	79.0	553	4.0405	0.81	0.2910	1.1716	0.81	0.7892	0.1451	0.0650
3.4085	80.0	560	4.0476	0.81	0.2917	1.1743	0.81	0.7892	0.1453	0.0650
3.4085	81.0	567	4.0529	0.805	0.2921	1.1736	0.805	0.7833	0.1573	0.0654
3.4085	82.0	574	4.0570	0.805	0.2919	1.1741	0.805	0.7861	0.1717	0.0655
3.4085	83.0	581	4.0601	0.81	0.2918	1.1727	0.81	0.7892	0.1508	0.0650
3.4085	84.0	588	4.0643	0.81	0.2919	1.1743	0.81	0.7892	0.1507	0.0652
3.4085	85.0	595	4.0678	0.81	0.2922	1.1744	0.81	0.7892	0.1552	0.0651
3.4085	86.0	602	4.0743	0.81	0.2925	1.1746	0.81	0.7892	0.1526	0.0651
3.4085	87.0	609	4.0758	0.805	0.2924	1.1753	0.805	0.7833	0.1718	0.0653
3.4085	88.0	616	4.0796	0.805	0.2924	1.1758	0.805	0.7833	0.1567	0.0654
3.4085	89.0	623	4.0803	0.81	0.2920	1.1742	0.81	0.7892	0.1587	0.0650
3.4085	90.0	630	4.0842	0.81	0.2925	1.1744	0.81	0.7892	0.1529	0.0651
3.4085	91.0	637	4.0864	0.805	0.2926	1.1752	0.805	0.7833	0.1568	0.0654
3.4085	92.0	644	4.0880	0.81	0.2925	1.1757	0.81	0.7892	0.1526	0.0651
3.4085	93.0	651	4.0903	0.805	0.2927	1.1752	0.805	0.7833	0.1567	0.0654
3.4085	94.0	658	4.0918	0.805	0.2927	1.1750	0.805	0.7833	0.1572	0.0655
3.4085	95.0	665	4.0927	0.805	0.2926	1.1750	0.805	0.7833	0.1570	0.0655
3.4085	96.0	672	4.0937	0.805	0.2927	1.1751	0.805	0.7833	0.1572	0.0655
3.4085	97.0	679	4.0946	0.805	0.2926	1.1750	0.805	0.7833	0.1573	0.0655
3.4085	98.0	686	4.0950	0.805	0.2926	1.1752	0.805	0.7833	0.1572	0.0655
3.4085	99.0	693	4.0955	0.805	0.2927	1.1753	0.805	0.7833	0.1572	0.0655
3.4085	100.0	700	4.0957	0.805	0.2927	1.1753	0.805	0.7833	0.1572	0.0655

Framework versions

Transformers 4.26.1
Pytorch 1.13.1.post200
Datasets 2.9.0
Tokenizers 0.13.2

jordyvl/18-tiny_tobacco3482_kd_NKD_t1.0_g1.5

18-tiny_tobacco3482_kd_NKD_t1.0_g1.5

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions