README.md

metadata

library_name: transformers
language:
  - kz
license: apache-2.0
base_model: facebook/wav2vec2-large-xlsr-53
tags:
  - hf-asr-leaderboard
  - generated_from_trainer
datasets:
  - Commonvoice-kazakh
metrics:
  - wer
model-index:
  - name: Kammi
    results:
      - task:
          type: automatic-speech-recognition
          name: Automatic Speech Recognition
        dataset:
          name: BilalS96/Commonvoice-kazakh
          type: Commonvoice-kazakh
          config: kk
          split: None
          args: 'config: kzk, split: test'
        metrics:
          - type: wer
            value: 1
            name: Wer

Kammi

This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on the BilalS96/Commonvoice-kazakh dataset. It achieves the following results on the evaluation set:

• Loss: 3.2424
• Wer: 1.0

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: 0.0003
• train_batch_size: 4
• eval_batch_size: 8
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
3.7719	4.3860	500	3.2418	1.0
3.2332	8.7719	1000	3.2473	1.0
3.2332	13.1579	1500	3.2480	1.0
3.227	17.5439	2000	3.2337	1.0
3.2218	21.9298	2500	3.2342	1.0
3.2361	26.3158	3000	3.2424	1.0

Framework versions

• Transformers 4.48.3
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0