xls-r-uyghur-cv8 / .ipynb_checkpoints /README-checkpoint.md
lucio's picture
Training in progress, step 17500
6b9b55b
|
raw
history blame
2.63 kB
metadata
language:
  - ug
license: apache-2.0
tags:
  - automatic-speech-recognition
  - mozilla-foundation/common_voice_8_0
  - generated_from_trainer
datasets:
  - common_voice
model-index:
  - name: xls-r-uyghur-cv8
    results: []

xls-r-uyghur-cv8

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the MOZILLA-FOUNDATION/COMMON_VOICE_8_0 - UG dataset. It achieves the following results on the evaluation set:

  • Loss: 0.2430
  • Wer: 0.3804

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: 7.5e-05
  • train_batch_size: 8
  • eval_batch_size: 8
  • seed: 42
  • gradient_accumulation_steps: 4
  • total_train_batch_size: 32
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 2000
  • num_epochs: 50.0
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
3.6871 2.66 500 3.5374 1.0
3.1501 5.32 1000 3.1278 1.0
1.5843 7.97 1500 0.6358 0.6914
1.3378 10.64 2000 0.4422 0.5925
1.2595 13.3 2500 0.3921 0.5512
1.1643 15.95 3000 0.3507 0.5149
1.1352 18.61 3500 0.3351 0.5019
1.1113 21.28 4000 0.3153 0.4845
1.0914 23.93 4500 0.3050 0.4594
1.0468 26.59 5000 0.2890 0.4470
1.0473 29.25 5500 0.2755 0.4331
1.0065 31.91 6000 0.2718 0.4264
0.9794 34.57 6500 0.2646 0.4193
0.9849 37.23 7000 0.2610 0.4058
0.9496 39.89 7500 0.2522 0.3985
0.9367 42.55 8000 0.2514 0.3947
0.9295 45.21 8500 0.2458 0.3883
0.9187 47.87 9000 0.2439 0.3833

Framework versions

  • Transformers 4.16.0.dev0
  • Pytorch 1.10.1+cu102
  • Datasets 1.18.2.dev0
  • Tokenizers 0.11.0