xwhisper-kh-small / README.md
PhanithLIM's picture
Update README.md
a4bc194 verified
metadata
license: apache-2.0
datasets:
  - openslr/openslr
  - google/fleurs
  - PhanithLIM/rfi-news-dataset
  - seanghay/km-speech-corpus
language:
  - km
metrics:
  - wer
base_model:
  - openai/whisper-small
pipeline_tag: automatic-speech-recognition
widget:
  - src: output/1.wav
    example_title: Audio 1
    output:
      text: >-
        ក្នុងរាត្រីកាលដ៏ស្ងប់ស្ងាត់មួយ
        បានផ្តិតជាប់នៅរូបភាពដ៏សែនសោកសង្រែងជាខ្លាំងចំពោះបុរសចំទង់ម៉ុនាស់
  - src: output/2.wav
    example_title: Audio 2
    output:
      text: ពុក កុំជាទៅដល់ហើយ!សុំទេវិត្តអាចពុកកុំអោយកើតឯងមុនពេលខ្ញុំទៅដល់!

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

  • eval_loss: 0.18
  • eval_wer: 65.4881 (0.654881)
  • eval_runtime: 2738.0001
  • eval_samples_per_second: 1.588
  • eval_steps_per_second: 0.199
  • epoch: 4.0
  • step: 4345

Training hyperparameters

The following hyperparameters were used during training:

  • learning_rate: 1e-05
  • train_batch_size: 16
  • eval_batch_size: 8
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: constant
  • lr_scheduler_warmup_steps: 1000
  • num_epochs: 10

Framework versions

  • Transformers 4.45.2
  • Pytorch 2.5.1+cu121
  • Datasets 3.1.0
  • Tokenizers 0.20.3