visual-emotion-recognition

This model is a fine-tuned version of google/vit-base-patch16-224-in21k on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 1.1334
• Accuracy: 0.6375
• Precision: 0.6498
• Recall: 0.6375
• F1: 0.6341

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: 1e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 3
• total_train_batch_size: 48
• 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
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
2.0671	0.97	13	2.0660	0.125	0.2709	0.125	0.1135
2.0576	1.95	26	2.0563	0.1562	0.2932	0.1562	0.1402
2.044	3.0	40	2.0439	0.1875	0.2554	0.1875	0.1827
2.0209	3.98	53	2.0309	0.2062	0.2405	0.2062	0.1961
1.9938	4.95	66	2.0176	0.2188	0.2410	0.2188	0.2062
1.9894	6.0	80	1.9960	0.2625	0.2700	0.2625	0.2438
1.9667	6.97	93	1.9743	0.3125	0.3089	0.3125	0.2901
1.9158	7.95	106	1.9421	0.3063	0.2557	0.3063	0.2687
1.8834	9.0	120	1.9042	0.3375	0.4019	0.3375	0.2888
1.8461	9.97	133	1.8521	0.3625	0.4132	0.3625	0.3021
1.7917	10.95	146	1.8023	0.3688	0.4144	0.3688	0.3056
1.7685	12.0	160	1.7552	0.375	0.4062	0.375	0.2978
1.7072	12.97	173	1.7071	0.3875	0.4266	0.3875	0.3164
1.6926	13.95	186	1.6742	0.375	0.4056	0.375	0.2996
1.6084	15.0	200	1.6476	0.3937	0.4411	0.3937	0.3358
1.6264	15.97	213	1.6231	0.3812	0.4357	0.3812	0.3311
1.5531	16.95	226	1.6019	0.4125	0.4676	0.4125	0.3626
1.5804	18.0	240	1.5773	0.3937	0.4442	0.3937	0.3428
1.54	18.98	253	1.5606	0.4	0.4565	0.4	0.3527
1.5461	19.95	266	1.5464	0.4437	0.5084	0.4437	0.4028
1.4841	21.0	280	1.5323	0.4313	0.4950	0.4313	0.3881
1.4765	21.98	293	1.5121	0.4313	0.4884	0.4313	0.3822
1.4838	22.95	306	1.4978	0.4375	0.5138	0.4375	0.4012
1.4487	24.0	320	1.4791	0.4437	0.5059	0.4437	0.4001
1.4272	24.98	333	1.4617	0.4562	0.5304	0.4562	0.4180
1.3886	25.95	346	1.4488	0.4625	0.5418	0.4625	0.4303
1.4529	27.0	360	1.4436	0.45	0.5147	0.45	0.4035
1.3894	27.98	373	1.4267	0.4688	0.5488	0.4688	0.4355
1.3848	28.95	386	1.4153	0.4625	0.5337	0.4625	0.4264
1.3561	30.0	400	1.3993	0.4875	0.5521	0.4875	0.4554
1.3184	30.98	413	1.3852	0.4813	0.5526	0.4813	0.4470
1.282	31.95	426	1.3703	0.4813	0.5480	0.4813	0.4449
1.2988	33.0	440	1.3674	0.4688	0.5541	0.4688	0.4395
1.2507	33.98	453	1.3594	0.4688	0.5347	0.4688	0.4307
1.2446	34.95	466	1.3519	0.4813	0.5616	0.4813	0.4514
1.2877	36.0	480	1.3547	0.4875	0.5599	0.4875	0.4605
1.2237	36.98	493	1.3342	0.5	0.5744	0.5	0.4654
1.2416	37.95	506	1.3214	0.4813	0.5693	0.4813	0.4551
1.1786	39.0	520	1.3122	0.4875	0.5674	0.4875	0.4586
1.193	39.98	533	1.2989	0.5	0.5755	0.5	0.4774
1.148	40.95	546	1.2962	0.5125	0.5811	0.5125	0.4755
1.1904	42.0	560	1.2860	0.5188	0.5863	0.5188	0.4928
1.1311	42.98	573	1.2893	0.5312	0.5936	0.5312	0.5117
1.1396	43.95	586	1.2860	0.4938	0.5633	0.4938	0.4698
1.1235	45.0	600	1.2802	0.5	0.5725	0.5	0.4758
1.1638	45.98	613	1.2596	0.525	0.5909	0.525	0.5058
1.0777	46.95	626	1.2668	0.5188	0.5796	0.5188	0.4861
1.1136	48.0	640	1.2520	0.55	0.6100	0.55	0.5291
1.047	48.98	653	1.2437	0.5375	0.5963	0.5375	0.5279
1.1101	49.95	666	1.2527	0.55	0.6195	0.55	0.5279
1.0412	51.0	680	1.2455	0.525	0.5927	0.525	0.5156
1.041	51.98	693	1.2245	0.55	0.6073	0.55	0.5353
0.9906	52.95	706	1.2307	0.575	0.6420	0.575	0.5600
0.9863	54.0	720	1.2307	0.5563	0.6150	0.5563	0.5362
0.943	54.98	733	1.2270	0.55	0.6152	0.55	0.5302
0.9557	55.95	746	1.2063	0.5312	0.5964	0.5312	0.5239
0.9518	57.0	760	1.2122	0.55	0.6232	0.55	0.5433
0.9545	57.98	773	1.1955	0.575	0.6144	0.575	0.5563
0.9195	58.95	786	1.2139	0.5563	0.6052	0.5563	0.5459
0.9267	60.0	800	1.1907	0.5687	0.6052	0.5687	0.5595
0.9384	60.98	813	1.1899	0.575	0.6449	0.575	0.5650
0.8727	61.95	826	1.1854	0.5813	0.6312	0.5813	0.5651
0.8541	63.0	840	1.1957	0.575	0.6407	0.575	0.5632
0.8899	63.98	853	1.1604	0.575	0.6196	0.575	0.5694
0.9036	64.95	866	1.1859	0.5563	0.6310	0.5563	0.5306
0.8177	66.0	880	1.1498	0.6125	0.6316	0.6125	0.6116
0.7854	66.97	893	1.1842	0.5687	0.6142	0.5687	0.5582
0.8054	67.95	906	1.1695	0.5938	0.6275	0.5938	0.5830
0.8582	69.0	920	1.1882	0.5687	0.6057	0.5687	0.5495
0.7603	69.97	933	1.2067	0.55	0.6025	0.55	0.5348
0.763	70.95	946	1.1690	0.5625	0.6036	0.5625	0.5439
0.8261	72.0	960	1.1616	0.6062	0.6306	0.6062	0.6016
0.884	72.97	973	1.1952	0.5625	0.6082	0.5625	0.5436
0.7843	73.95	986	1.1583	0.5687	0.5953	0.5687	0.5633
0.801	75.0	1000	1.1547	0.575	0.6013	0.575	0.5745
0.7454	75.97	1013	1.1372	0.5875	0.6193	0.5875	0.5761
0.7325	76.95	1026	1.1696	0.5938	0.6351	0.5938	0.5919
0.7931	78.0	1040	1.1511	0.6062	0.6342	0.6062	0.6053
0.7487	78.97	1053	1.1655	0.5625	0.5898	0.5625	0.5496
0.7262	79.95	1066	1.1394	0.6125	0.6295	0.6125	0.6048
0.7669	81.0	1080	1.1748	0.575	0.5966	0.575	0.5697
0.7028	81.97	1093	1.1418	0.5875	0.6178	0.5875	0.5885
0.7749	82.95	1106	1.1736	0.55	0.5446	0.55	0.5255
0.7233	84.0	1120	1.1645	0.5813	0.5973	0.5813	0.5699
0.5915	84.97	1133	1.1376	0.5875	0.6167	0.5875	0.5867
0.6985	85.95	1146	1.1665	0.5687	0.5868	0.5687	0.5533
0.6572	87.0	1160	1.1341	0.6	0.6245	0.6	0.5963
0.6317	87.97	1173	1.1327	0.6125	0.6288	0.6125	0.6026
0.6546	88.95	1186	1.1668	0.5687	0.5797	0.5687	0.5528
0.5801	90.0	1200	1.1521	0.5875	0.6161	0.5875	0.5818
0.6958	90.97	1213	1.1401	0.5875	0.6083	0.5875	0.5774
0.5856	91.95	1226	1.1379	0.5875	0.5888	0.5875	0.5760
0.6281	93.0	1240	1.1379	0.6125	0.6429	0.6125	0.6123
0.6518	93.97	1253	1.1619	0.6312	0.6547	0.6312	0.6247
0.6055	94.95	1266	1.1700	0.575	0.5962	0.575	0.5673
0.6181	96.0	1280	1.1550	0.5938	0.6281	0.5938	0.5970
0.6601	96.97	1293	1.1334	0.6375	0.6498	0.6375	0.6341
0.6112	97.5	1300	1.1007	0.6188	0.6341	0.6188	0.6207

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.0+cu121
• Datasets 2.17.0
• Tokenizers 0.15.1