Hugging Face's logo

mizoru
/
whisper-small-ru-ORD_0.9_peft_0.3

PEFT
Safetensors
Russian
Generated from Trainer
Model card Files Community
whisper-small-ru-ORD_0.9_peft_0.3 / README.md
mizoru's picture
mizoru
End of training
ad85bc0 verified
preview code
|
raw
history blame contribute delete
2.4 kB
metadata
language:
  - ru
license: apache-2.0
library_name: peft
tags:
  - generated_from_trainer
base_model: openai/whisper-small
metrics:
  - wer
model-index:
  - name: 'Whisper Small Ru ORD 0.9 Peft PEFT 4-bit Q DoRA - Mizoru '
    results: []

Visualize in Weights & Biases Visualize in Weights & Biases

Whisper Small Ru ORD 0.9 Peft PEFT 4-bit Q DoRA - Mizoru

This model is a fine-tuned version of openai/whisper-small on the ORD_0.9 dataset. It achieves the following results on the evaluation set:

  • Loss: 1.1643
  • Wer: 58.1771
  • Cer: 31.9056
  • Clean Wer: 50.6879
  • Clean Cer: 26.1504

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: 16
  • eval_batch_size: 16
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 50
  • num_epochs: 4
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer Cer Clean Wer Clean Cer
1.1675 1.0 550 1.2152 60.5918 33.7819 55.0018 28.7141
1.1217 2.0 1100 1.1698 62.6194 35.1450 54.1401 29.5194
0.9579 3.0 1650 1.1557 58.2105 32.0513 51.0548 26.5161
0.7957 4.0 2200 1.1643 58.1771 31.9056 50.6879 26.1504

Framework versions

  • PEFT 0.11.1.dev0
  • Transformers 4.41.0.dev0
  • Pytorch 2.1.2
  • Datasets 2.18.0
  • Tokenizers 0.19.1