csikasote's picture
update model card README.md
aa74183
metadata
license: cc-by-nc-4.0
base_model: facebook/mms-1b-all
tags:
  - generated_from_trainer
metrics:
  - wer
model-index:
  - name: wav2vec2-large-mms-1b-nya-colab
    results: []

wav2vec2-large-mms-1b-nya-colab

This model is a fine-tuned version of facebook/mms-1b-all on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 0.4327
  • Wer: 0.3505

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: 0.001
  • train_batch_size: 4
  • eval_batch_size: 8
  • seed: 42
  • gradient_accumulation_steps: 2
  • total_train_batch_size: 8
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 100
  • num_epochs: 15

Training results

Training Loss Epoch Step Validation Loss Wer
4.1659 0.2 200 0.6822 0.5353
0.2331 0.39 400 0.5220 0.4493
0.2119 0.59 600 0.4967 0.4146
0.1995 0.79 800 0.5021 0.4025
0.1812 0.99 1000 0.5046 0.3979
0.1744 1.18 1200 0.4786 0.3884
0.1783 1.38 1400 0.4630 0.3786
0.1663 1.58 1600 0.4511 0.3634
0.1609 1.77 1800 0.4656 0.3647
0.1632 1.97 2000 0.4254 0.3553
0.1568 2.17 2200 0.4326 0.3529
0.1544 2.37 2400 0.4291 0.3477
0.1524 2.56 2600 0.4327 0.3505

Framework versions

  • Transformers 4.31.0.dev0
  • Pytorch 2.0.1+cu118
  • Datasets 2.13.1
  • Tokenizers 0.13.3