18-tiny_tobacco3482_hint_

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: 71.6043
Accuracy: 0.85
Brier Loss: 0.2638
Nll: 1.3122
F1 Micro: 0.85
F1 Macro: 0.8425
Ece: 0.1342
Aurc: 0.0355

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: 32
eval_batch_size: 32
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	25	74.9320	0.26	0.8713	4.9021	0.26	0.1920	0.3072	0.7822
No log	2.0	50	74.1171	0.54	0.5982	2.7516	0.54	0.4539	0.2763	0.2591
No log	3.0	75	73.6525	0.685	0.4431	1.5999	0.685	0.6196	0.1890	0.1355
No log	4.0	100	73.3675	0.735	0.3894	1.5393	0.735	0.7201	0.2334	0.1214
No log	5.0	125	73.0199	0.785	0.3130	1.6799	0.785	0.7426	0.1738	0.0705
No log	6.0	150	72.7837	0.8	0.3068	1.5290	0.8000	0.7711	0.1531	0.0582
No log	7.0	175	72.7218	0.76	0.3913	1.8944	0.76	0.7334	0.1998	0.0987
No log	8.0	200	72.7351	0.695	0.4805	1.4948	0.695	0.6565	0.2585	0.1329
No log	9.0	225	72.2520	0.79	0.3279	1.5198	0.79	0.7440	0.1878	0.0704
No log	10.0	250	72.3792	0.785	0.3370	1.7982	0.785	0.7687	0.1655	0.0757
No log	11.0	275	72.2408	0.765	0.3542	1.6304	0.765	0.7364	0.1838	0.0660
No log	12.0	300	72.1003	0.76	0.3664	1.4090	0.76	0.7247	0.1832	0.0817
No log	13.0	325	72.4671	0.77	0.3886	1.7820	0.7700	0.7380	0.1985	0.0833
No log	14.0	350	72.1519	0.785	0.3633	1.5939	0.785	0.7563	0.1814	0.0687
No log	15.0	375	72.2206	0.775	0.3723	1.6025	0.775	0.7406	0.1824	0.0800
No log	16.0	400	72.0554	0.805	0.3080	1.4448	0.805	0.7520	0.1577	0.0485
No log	17.0	425	72.3130	0.8	0.3412	1.7255	0.8000	0.7653	0.1637	0.0694
No log	18.0	450	72.1019	0.815	0.3228	1.4827	0.815	0.7918	0.1753	0.0592
No log	19.0	475	72.2643	0.8	0.3558	1.5960	0.8000	0.7604	0.1677	0.0737
71.4928	20.0	500	71.9509	0.78	0.3398	1.5615	0.78	0.7596	0.1683	0.0561
71.4928	21.0	525	71.9389	0.82	0.2947	1.5336	0.82	0.8008	0.1561	0.0538
71.4928	22.0	550	72.0399	0.83	0.2843	1.4576	0.83	0.7914	0.1347	0.0523
71.4928	23.0	575	72.0529	0.815	0.3263	1.3174	0.815	0.7923	0.1677	0.0523
71.4928	24.0	600	72.3487	0.775	0.3838	1.5418	0.775	0.7560	0.1880	0.0794
71.4928	25.0	625	71.9154	0.825	0.2949	1.2538	0.825	0.7992	0.1471	0.0584
71.4928	26.0	650	72.0222	0.815	0.3151	1.5809	0.815	0.7830	0.1601	0.0594
71.4928	27.0	675	72.0422	0.815	0.3269	1.5161	0.815	0.7954	0.1606	0.0597
71.4928	28.0	700	72.0172	0.845	0.2828	1.3388	0.845	0.8350	0.1447	0.0649
71.4928	29.0	725	71.9113	0.84	0.2685	1.2082	0.8400	0.8202	0.1365	0.0562
71.4928	30.0	750	71.9516	0.84	0.2856	1.2664	0.8400	0.8359	0.1415	0.0563
71.4928	31.0	775	71.8583	0.835	0.2782	1.2979	0.835	0.8277	0.1447	0.0545
71.4928	32.0	800	71.9071	0.84	0.2766	1.2772	0.8400	0.8359	0.1378	0.0546
71.4928	33.0	825	71.8580	0.85	0.2699	1.2985	0.85	0.8482	0.1351	0.0525
71.4928	34.0	850	71.8499	0.835	0.2872	1.3022	0.835	0.8292	0.1462	0.0532
71.4928	35.0	875	72.0085	0.84	0.2897	1.3042	0.8400	0.8323	0.1420	0.0616
71.4928	36.0	900	71.8423	0.82	0.2929	1.2266	0.82	0.8056	0.1543	0.0521
71.4928	37.0	925	71.7886	0.845	0.2807	1.2181	0.845	0.8254	0.1332	0.0400
71.4928	38.0	950	71.8857	0.83	0.2877	1.4036	0.83	0.8166	0.1490	0.0480
71.4928	39.0	975	71.9388	0.83	0.2877	1.3374	0.83	0.8119	0.1459	0.0451
70.1673	40.0	1000	72.0368	0.8	0.3368	1.7112	0.8000	0.7897	0.1741	0.0578
70.1673	41.0	1025	72.0295	0.8	0.3208	1.5473	0.8000	0.7862	0.1587	0.0622
70.1673	42.0	1050	71.7048	0.86	0.2547	1.3240	0.8600	0.8374	0.1257	0.0408
70.1673	43.0	1075	71.7541	0.835	0.2680	1.4095	0.835	0.8178	0.1418	0.0445
70.1673	44.0	1100	71.7746	0.845	0.2721	1.5529	0.845	0.8383	0.1412	0.0405
70.1673	45.0	1125	71.7661	0.83	0.2908	1.5315	0.83	0.8104	0.1415	0.0437
70.1673	46.0	1150	71.7563	0.84	0.2787	1.4088	0.8400	0.8238	0.1360	0.0416
70.1673	47.0	1175	71.7670	0.84	0.2709	1.2801	0.8400	0.8303	0.1322	0.0401
70.1673	48.0	1200	71.7458	0.84	0.2699	1.4180	0.8400	0.8265	0.1433	0.0397
70.1673	49.0	1225	71.7226	0.84	0.2653	1.4126	0.8400	0.8265	0.1282	0.0390
70.1673	50.0	1250	71.7163	0.85	0.2608	1.4227	0.85	0.8402	0.1339	0.0394
70.1673	51.0	1275	71.7044	0.845	0.2612	1.4130	0.845	0.8371	0.1314	0.0387
70.1673	52.0	1300	71.6821	0.85	0.2545	1.4880	0.85	0.8392	0.1286	0.0383
70.1673	53.0	1325	71.6764	0.845	0.2598	1.3776	0.845	0.8301	0.1299	0.0377
70.1673	54.0	1350	71.6750	0.855	0.2590	1.3404	0.855	0.8479	0.1361	0.0383
70.1673	55.0	1375	71.7192	0.855	0.2543	1.4804	0.855	0.8482	0.1346	0.0388
70.1673	56.0	1400	71.6907	0.85	0.2552	1.4102	0.85	0.8389	0.1274	0.0377
70.1673	57.0	1425	71.6778	0.85	0.2572	1.4026	0.85	0.8392	0.1319	0.0389
70.1673	58.0	1450	71.6735	0.85	0.2559	1.4062	0.85	0.8394	0.1288	0.0386
70.1673	59.0	1475	71.6938	0.855	0.2549	1.4710	0.855	0.8482	0.1309	0.0387
69.9715	60.0	1500	71.6799	0.85	0.2599	1.3989	0.85	0.8389	0.1309	0.0384
69.9715	61.0	1525	71.6753	0.865	0.2539	1.3555	0.865	0.8619	0.1201	0.0374
69.9715	62.0	1550	71.6657	0.855	0.2562	1.4005	0.855	0.8453	0.1301	0.0365
69.9715	63.0	1575	71.6941	0.855	0.2569	1.4083	0.855	0.8453	0.1297	0.0369
69.9715	64.0	1600	71.6430	0.85	0.2567	1.3933	0.85	0.8395	0.1239	0.0370
69.9715	65.0	1625	71.6666	0.85	0.2582	1.4014	0.85	0.8395	0.1357	0.0375
69.9715	66.0	1650	71.6550	0.85	0.2578	1.3849	0.85	0.8395	0.1253	0.0370
69.9715	67.0	1675	71.6321	0.855	0.2573	1.3932	0.855	0.8466	0.1276	0.0362
69.9715	68.0	1700	71.6237	0.855	0.2576	1.3976	0.855	0.8453	0.1231	0.0374
69.9715	69.0	1725	71.6287	0.85	0.2589	1.3914	0.85	0.8403	0.1299	0.0366
69.9715	70.0	1750	71.6325	0.85	0.2580	1.3907	0.85	0.8425	0.1321	0.0365
69.9715	71.0	1775	71.6175	0.85	0.2572	1.3914	0.85	0.8412	0.1318	0.0365
69.9715	72.0	1800	71.6208	0.85	0.2591	1.3860	0.85	0.8425	0.1325	0.0355
69.9715	73.0	1825	71.6157	0.85	0.2600	1.3894	0.85	0.8425	0.1335	0.0361
69.9715	74.0	1850	71.6405	0.85	0.2632	1.3335	0.85	0.8425	0.1306	0.0359
69.9715	75.0	1875	71.6099	0.85	0.2586	1.3899	0.85	0.8425	0.1283	0.0360
69.9715	76.0	1900	71.6058	0.85	0.2599	1.3220	0.85	0.8425	0.1260	0.0357
69.9715	77.0	1925	71.6096	0.85	0.2591	1.3859	0.85	0.8425	0.1279	0.0355
69.9715	78.0	1950	71.6213	0.85	0.2604	1.3875	0.85	0.8425	0.1284	0.0351
69.9715	79.0	1975	71.6240	0.85	0.2610	1.3867	0.85	0.8425	0.1347	0.0356
69.8814	80.0	2000	71.6246	0.855	0.2608	1.3183	0.855	0.8483	0.1238	0.0354
69.8814	81.0	2025	71.6133	0.85	0.2595	1.3177	0.85	0.8425	0.1268	0.0357
69.8814	82.0	2050	71.6120	0.85	0.2593	1.3873	0.85	0.8425	0.1278	0.0353
69.8814	83.0	2075	71.6183	0.85	0.2600	1.3220	0.85	0.8425	0.1342	0.0356
69.8814	84.0	2100	71.6008	0.85	0.2610	1.3168	0.85	0.8425	0.1359	0.0358
69.8814	85.0	2125	71.5986	0.85	0.2621	1.3156	0.85	0.8425	0.1327	0.0355
69.8814	86.0	2150	71.6000	0.85	0.2609	1.3167	0.85	0.8425	0.1311	0.0354
69.8814	87.0	2175	71.6006	0.855	0.2611	1.3171	0.855	0.8483	0.1251	0.0353
69.8814	88.0	2200	71.6033	0.85	0.2620	1.3132	0.85	0.8425	0.1331	0.0357
69.8814	89.0	2225	71.6176	0.855	0.2635	1.3142	0.855	0.8483	0.1326	0.0357
69.8814	90.0	2250	71.6201	0.855	0.2636	1.3126	0.855	0.8483	0.1282	0.0356
69.8814	91.0	2275	71.6128	0.85	0.2629	1.3126	0.85	0.8425	0.1273	0.0356
69.8814	92.0	2300	71.6086	0.855	0.2631	1.3147	0.855	0.8515	0.1261	0.0358
69.8814	93.0	2325	71.6010	0.85	0.2638	1.3122	0.85	0.8425	0.1292	0.0356
69.8814	94.0	2350	71.6053	0.85	0.2636	1.3125	0.85	0.8425	0.1269	0.0354
69.8814	95.0	2375	71.6004	0.85	0.2640	1.3128	0.85	0.8425	0.1346	0.0356
69.8814	96.0	2400	71.6035	0.85	0.2644	1.3128	0.85	0.8425	0.1346	0.0356
69.8814	97.0	2425	71.6027	0.85	0.2639	1.3117	0.85	0.8425	0.1343	0.0355
69.8814	98.0	2450	71.6039	0.85	0.2639	1.3117	0.85	0.8425	0.1343	0.0354
69.8814	99.0	2475	71.6018	0.85	0.2640	1.3122	0.85	0.8425	0.1342	0.0356
69.8448	100.0	2500	71.6043	0.85	0.2638	1.3122	0.85	0.8425	0.1342	0.0355

Framework versions

Transformers 4.26.1
Pytorch 1.13.1.post200
Datasets 2.9.0
Tokenizers 0.13.2

jordyvl/18-tiny_tobacco3482_hint_

18-tiny_tobacco3482_hint_

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions