bryan6aero's picture
update model card README.md
7b51543
metadata
license: apache-2.0
tags:
  - generated_from_trainer
model-index:
  - name: wav2vec2-base-timit-demo-colab
    results: []

wav2vec2-base-timit-demo-colab

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: 0.4779
  • Wer: 0.3453

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: 32
  • eval_batch_size: 8
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 1000
  • num_epochs: 30
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
3.4307 4.0 500 1.4129 0.9980
0.626 8.0 1000 0.4605 0.4499
0.2199 12.0 1500 0.4457 0.3898
0.1303 16.0 2000 0.4418 0.3771
0.0851 20.0 2500 0.4647 0.3548
0.0604 24.0 3000 0.4603 0.3499
0.0461 28.0 3500 0.4779 0.3453

Framework versions

  • Transformers 4.11.3
  • Pytorch 1.10.0+cu111
  • Datasets 1.18.3
  • Tokenizers 0.10.3