PhanithLIM's picture
Update README.md
cdae1fd verified
metadata
library_name: transformers
license: apache-2.0
base_model: openai/whisper-tiny
datasets:
  - PhanithLIM/ams-speech-dataset
  - openslr/openslr
  - google/fleurs
  - PhanithLIM/kh-wmc
  - PhanithLIM/wmc-international-news
  - PhanithLIM/rfi-news-dataset
  - PhanithLIM/aakanee-kh
  - rinabuoy/khm-asr-open
  - seanghay/khmer_grkpp_speech
  - seanghay/khmer_mpwt_speech
  - seanghay/km-speech-corpus
model-index:
  - name: Khmer Whisper Small PhanithLIM
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: Google Fleurs
          type: google/fleurs
          config: km_kh
          split: test
        metrics:
          - name: CER
            type: cer
            value: 22.511
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: WMC
          type: PhanithLIM/asr-wmc-evaluate
          split: test
        metrics:
          - name: CER
            type: cer
            value: 12.581
tags:
  - generated_from_trainer
metrics:
  - wer

whisper-tiny-aug-7-may-lightning-v1

This model is a fine-tuned version of openai/whisper-tiny on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.1300
  • Wer: 86.2590

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: 1e-05
  • train_batch_size: 32
  • eval_batch_size: 32
  • seed: 42
  • gradient_accumulation_steps: 2
  • total_train_batch_size: 64
  • optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
  • lr_scheduler_type: constant
  • lr_scheduler_warmup_steps: 1000
  • num_epochs: 10

Training results

Training Loss Epoch Step Validation Loss Wer
1.0747 1.0 712 0.4463 102.0236
0.3496 2.0 1424 0.2607 98.4686
0.2411 3.0 2136 0.2071 92.8878
0.1966 4.0 2848 0.1819 94.1085
0.1699 5.0 3560 0.1653 92.2555
0.1514 6.0 4272 0.1533 88.5561
0.1377 7.0 4984 0.1452 88.0289
0.1265 8.0 5696 0.1391 86.8913
0.117 9.0 6408 0.1331 87.4382
0.1089 10.0 7120 0.1300 86.2590

Framework versions

  • Transformers 4.51.3
  • Pytorch 2.7.0+cu128
  • Datasets 3.5.1
  • Tokenizers 0.21.1