13E-affecthq-fer-balanced

This model is a fine-tuned version of google/vit-base-patch16-224-in21k on Piro17/balancednumber-affecthqnet-fer2013 dataset. It achieves the following results on the evaluation set:

• Loss: 1.0526
• Accuracy: 0.6225
• Precision: 0.6161
• Recall: 0.6225
• F1: 0.6167

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: 32
• eval_batch_size: 32
• seed: 17
• gradient_accumulation_steps: 4
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 13

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
1.7863	1.0	133	1.7632	0.4005	0.3617	0.4005	0.3058
1.3653	2.0	266	1.3630	0.5049	0.4838	0.5049	0.4445
1.2468	3.0	399	1.2475	0.5466	0.5451	0.5466	0.5115
1.1527	4.0	532	1.1865	0.5761	0.5612	0.5761	0.5580
1.0862	5.0	665	1.1448	0.5785	0.5687	0.5785	0.5659
1.064	6.0	798	1.1108	0.5972	0.5867	0.5972	0.5853
1.0037	7.0	931	1.0969	0.6019	0.5968	0.6019	0.5946
0.9533	8.0	1064	1.0764	0.6126	0.6034	0.6126	0.6046
0.9063	9.0	1197	1.0711	0.6155	0.6035	0.6155	0.6047
0.8666	10.0	1330	1.0589	0.6173	0.6107	0.6173	0.6108
0.8364	11.0	1463	1.0556	0.6178	0.6110	0.6178	0.6108
0.8659	12.0	1596	1.0521	0.6197	0.6141	0.6197	0.6151
0.8383	13.0	1729	1.0526	0.6225	0.6161	0.6225	0.6167

Framework versions

• Transformers 4.27.0.dev0
• Pytorch 1.13.1+cu116
• Datasets 2.9.0
• Tokenizers 0.13.2