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: 1.3469
• Accuracy: 0.175

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
No log	1.0	10	2.0721	0.125
No log	2.0	20	2.0633	0.125
No log	3.0	30	2.0038	0.125
No log	4.0	40	1.9097	0.125
No log	5.0	50	1.7412	0.125
No log	6.0	60	1.6189	0.05
No log	7.0	70	1.5343	0.0375
No log	8.0	80	1.4746	0.0688
No log	9.0	90	1.4330	0.0938
No log	10.0	100	1.4130	0.15
No log	11.0	110	1.3735	0.1062
No log	12.0	120	1.3516	0.1062
No log	13.0	130	1.2838	0.1375
No log	14.0	140	1.3058	0.1187
No log	15.0	150	1.3116	0.1
No log	16.0	160	1.3269	0.1313
No log	17.0	170	1.2624	0.1062
No log	18.0	180	1.3285	0.1187
No log	19.0	190	1.3490	0.1437
No log	20.0	200	1.2592	0.1375
No log	21.0	210	1.3600	0.0938
No log	22.0	220	1.2835	0.1313
No log	23.0	230	1.2842	0.1375
No log	24.0	240	1.2840	0.1
No log	25.0	250	1.2456	0.1313
No log	26.0	260	1.2960	0.1562
No log	27.0	270	1.3208	0.1375
No log	28.0	280	1.3207	0.1375
No log	29.0	290	1.2892	0.175
No log	30.0	300	1.2837	0.1812
No log	31.0	310	1.3548	0.1562
No log	32.0	320	1.4371	0.1437
No log	33.0	330	1.4219	0.1562
No log	34.0	340	1.4033	0.1875
No log	35.0	350	1.4505	0.1437
No log	36.0	360	1.2975	0.1562
No log	37.0	370	1.3906	0.1562
No log	38.0	380	1.3547	0.1688
No log	39.0	390	1.4706	0.1938
No log	40.0	400	1.3595	0.1625
No log	41.0	410	1.4236	0.1625
No log	42.0	420	1.4180	0.1812
No log	43.0	430	1.3993	0.1562
No log	44.0	440	1.4066	0.1625
No log	45.0	450	1.3760	0.175
No log	46.0	460	1.4221	0.1812
No log	47.0	470	1.3772	0.1625
No log	48.0	480	1.4265	0.2
No log	49.0	490	1.4716	0.1625
0.6962	50.0	500	1.3917	0.1625

Framework versions

• Transformers 4.33.1
• Pytorch 2.0.1+cu117
• Datasets 2.14.5
• Tokenizers 0.13.3