argish/wav2vec2-base-960h-speech-emotion-classification-E02_SER

wav2vec2-base-960h_SER_merged_dataset

This model is a fine-tuned version of facebook/wav2vec2-base-960h on a merged dataset of RAVDESS, CREMA, SAVEE, and TESS. It achieves the following results on the evaluation set:

• eval_accuracy: 0.7480
• eval_loss: 0.7940
• eval_weighted_f1: 0.7438
• eval_micro_f1: 0.7480
• eval_macro_f1: 0.7559
• eval_weighted_recall: 0.7480
• eval_micro_recall: 0.7480
• eval_macro_recall: 0.7622
• eval_weighted_precision: 0.7518
• eval_micro_precision: 0.7480
• eval_macro_precision: 0.7609
• eval_runtime: 82.1913
• eval_samples_per_second: 29.602
• eval_steps_per_second: 0.937
• epoch: 14.0
• step: 1078

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: 3e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 128
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 15

Framework versions

• Transformers 4.47.0
• Pytorch 2.5.1+cu121
• Datasets 3.3.1
• Tokenizers 0.21.0