vit-emotion-classifier

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: 1.3090
• Accuracy: 0.55

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: 5e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.4729	1.0	10	1.5748	0.4875
1.4484	2.0	20	1.5526	0.4875
1.4053	3.0	30	1.5228	0.4562
1.3492	4.0	40	1.4721	0.5
1.2664	5.0	50	1.4448	0.5125
1.2005	6.0	60	1.3783	0.5062
1.1231	7.0	70	1.3427	0.5375
1.0472	8.0	80	1.2859	0.5625
0.9852	9.0	90	1.2732	0.5813
0.8974	10.0	100	1.2220	0.575
0.8314	11.0	110	1.2782	0.5312
0.7964	12.0	120	1.2889	0.5437
0.6993	13.0	130	1.2989	0.5188
0.6915	14.0	140	1.3053	0.5375
0.608	15.0	150	1.2563	0.5875
0.5416	16.0	160	1.2473	0.5563
0.5202	17.0	170	1.2753	0.5625
0.5047	18.0	180	1.2791	0.5563
0.4779	19.0	190	1.3142	0.5437
0.4569	20.0	200	1.2743	0.5813
0.4313	21.0	210	1.2727	0.5312
0.4536	22.0	220	1.2514	0.5938
0.4166	23.0	230	1.3260	0.5312
0.3673	24.0	240	1.2950	0.55
0.3544	25.0	250	1.2268	0.5875
0.3568	26.0	260	1.3874	0.4875
0.3509	27.0	270	1.3735	0.525
0.3711	28.0	280	1.2886	0.5375
0.3555	29.0	290	1.3152	0.5375
0.3068	30.0	300	1.3927	0.5375
0.3007	31.0	310	1.4131	0.5188
0.3062	32.0	320	1.3256	0.575
0.3114	33.0	330	1.3714	0.5
0.279	34.0	340	1.4198	0.5188
0.2888	35.0	350	1.5321	0.475
0.2647	36.0	360	1.4342	0.5062
0.2574	37.0	370	1.4149	0.5563
0.2539	38.0	380	1.4286	0.5125
0.2566	39.0	390	1.4805	0.5125
0.2298	40.0	400	1.3820	0.4875
0.2236	41.0	410	1.3683	0.5437
0.2201	42.0	420	1.3332	0.5687
0.2696	43.0	430	1.4725	0.5188
0.2319	44.0	440	1.3926	0.5375
0.2269	45.0	450	1.3477	0.5563
0.2201	46.0	460	1.4054	0.5563
0.2114	47.0	470	1.3308	0.55
0.2319	48.0	480	1.3353	0.5625
0.2177	49.0	490	1.3019	0.5437
0.2042	50.0	500	1.3089	0.5875

Framework versions

• Transformers 4.33.2
• Pytorch 2.0.1+cu118
• Datasets 2.14.5
• Tokenizers 0.13.3