jethrowang's picture
End of training
2f95141 verified
|
raw
history blame
2.72 kB
metadata
language:
  - zh
license: apache-2.0
base_model: openai/whisper-tiny
tags:
  - generated_from_trainer
datasets:
  - formospeech/hat_asr_aligned
model-index:
  - name: Whisper Tiny Hakka Simulated Webcam
    results: []

Whisper Tiny Hakka Simulated Webcam

This model is a fine-tuned version of openai/whisper-tiny on the HAT ASR Aligned dataset. It achieves the following results on the evaluation set:

  • Loss: 0.1959
  • Cer: 12.2188

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: 64
  • eval_batch_size: 32
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 976
  • training_steps: 9760
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Cer
1.189 0.9980 488 1.2025 50.1503
0.3904 1.9959 976 0.4830 26.8916
0.2027 2.9939 1464 0.3017 17.2273
0.1241 3.9918 1952 0.2566 15.3859
0.0837 4.9898 2440 0.2299 14.5098
0.0558 5.9877 2928 0.2175 13.6302
0.0365 6.9857 3416 0.2119 13.6151
0.0266 7.9836 3904 0.2052 13.6059
0.0197 8.9816 4392 0.1990 11.9877
0.0131 9.9796 4880 0.1982 12.7887
0.0082 10.9775 5368 0.1987 12.5864
0.006 11.9755 5856 0.1985 13.6336
0.0046 12.9734 6344 0.1971 13.0037
0.0035 13.9714 6832 0.1945 12.7390
0.0034 14.9693 7320 0.1966 12.7135
0.0026 15.9673 7808 0.1954 12.6477
0.0022 16.9652 8296 0.1958 12.5922
0.0021 17.9632 8784 0.1957 11.5970
0.0019 18.9611 9272 0.1959 12.0061
0.0018 19.9591 9760 0.1959 12.2188

Framework versions

  • Transformers 4.42.3
  • Pytorch 2.3.0+cu121
  • Datasets 2.20.0
  • Tokenizers 0.19.1