Valence-wav2vec2-base-EMOPIA

This model is a fine-tuned version of facebook/wav2vec2-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.1044
• Accuracy: 0.6761

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-06
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• gradient_accumulation_steps: 3
• total_train_batch_size: 3
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 1000
• num_epochs: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.6979	1.0	269	0.6842	0.5915
0.69	2.0	538	0.6864	0.5775
0.6714	3.0	807	0.6900	0.5070
0.6357	4.0	1076	0.6514	0.5775
0.5678	5.0	1345	0.6612	0.6197
0.5152	6.0	1614	0.6496	0.6761
0.4826	7.0	1883	0.7743	0.6479
0.4707	8.0	2152	0.8348	0.6620
0.4742	9.0	2421	0.8808	0.6761
0.4857	10.0	2690	0.8734	0.7324
0.4779	11.0	2959	1.0206	0.6620
0.5063	12.0	3228	1.0737	0.6761
0.4776	13.0	3497	1.0966	0.6761
0.4716	14.0	3766	1.1046	0.6761
0.4672	15.0	4035	1.1044	0.6761

Framework versions

• Transformers 4.37.2
• Pytorch 2.1.0+cu121
• Datasets 2.16.1
• Tokenizers 0.15.1