MelanieKoe's picture
Model save
1d067ec verified
|
raw
history blame
2.37 kB
metadata
license: apache-2.0
base_model: facebook/wav2vec2-base
tags:
  - generated_from_trainer
metrics:
  - wer
model-index:
  - name: w2v2-base-pretrained_lr5e-5_at0.9_da1
    results: []

w2v2-base-pretrained_lr5e-5_at0.9_da1

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: 2.0012
  • Wer: 0.1794

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: 5e-05
  • 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: 500
  • training_steps: 4000
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
15.834 6.1 250 3.6941 1.0
3.1623 12.2 500 3.3447 1.0
2.81 18.29 750 1.9439 0.9983
0.4575 24.39 1000 1.2543 0.2499
0.1302 30.49 1250 1.5064 0.2080
0.0852 36.59 1500 1.5567 0.1995
0.0618 42.68 1750 1.8019 0.1982
0.0497 48.78 2000 1.8515 0.2025
0.0395 54.88 2250 1.8758 0.1897
0.0301 60.98 2500 1.7991 0.1850
0.0275 67.07 2750 1.9686 0.1777
0.0213 73.17 3000 1.8964 0.1884
0.02 79.27 3250 1.9815 0.1854
0.017 85.37 3500 2.0240 0.1790
0.0157 91.46 3750 1.9606 0.1768
0.0137 97.56 4000 2.0012 0.1794

Framework versions

  • Transformers 4.35.0
  • Pytorch 2.0.0
  • Datasets 2.14.6
  • Tokenizers 0.14.1