Edit model card

wav2vec2-base-finetuned-spgi-speech-dev

This model is a fine-tuned version of facebook/wav2vec2-base on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.2536
  • Wer: 0.1392

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: 0.0001
  • train_batch_size: 16
  • eval_batch_size: 8
  • 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_steps: 1000
  • num_epochs: 40
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
1.3183 6.8 2000 0.2018 0.1813
0.1081 13.6 4000 0.2168 0.1571
0.07 20.41 6000 0.2324 0.1510
0.0519 27.21 8000 0.2336 0.1414
0.0418 34.01 10000 0.2536 0.1392

Framework versions

  • Transformers 4.21.2
  • Pytorch 1.12.1+cu113
  • Datasets 2.4.0
  • Tokenizers 0.12.1
Downloads last month
1
Hosted inference API
or or
This model can be loaded on the Inference API on-demand.