---
language:
- ga
- en
license: apache-2.0
base_model: openai/whisper-small
tags:
- generated_from_trainer
datasets:
- ymoslem/IWSLT2023-GA-EN
- ymoslem/FLEURS-GA-EN
- ymoslem/BitesizeIrish-GA-EN
- ymoslem/SpokenWords-GA-EN-MTed
metrics:
- bleu
- wer
model-index:
- name: Whisper Small GA-EN Speech Translation
  results:
  - task:
      name: Automatic Speech Recognition
      type: automatic-speech-recognition
    dataset:
      name: IWSLT-2023, FLEURS, BiteSize, and SpokenWords
      type: ymoslem/IWSLT2023-GA-EN
    metrics:
    - name: Bleu
      type: bleu
      value: 29.51
    - name: Wer
      type: wer
      value: 67.08689779378658
---

<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->

# Whisper Small GA-EN Speech Translation

This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co./openai/whisper-small) on the IWSLT-2023, FLEURS, BiteSize, and SpokenWords dataset.
It achieves the following results on the evaluation set:
- Loss: 1.8231
- Bleu: 29.51
- Chrf: 44.29
- Wer: 67.0869

## 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: 0.0001
- train_batch_size: 64
- eval_batch_size: 64
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- training_steps: 4000
- mixed_precision_training: Native AMP

### Training results

| Training Loss | Epoch  | Step | Bleu  | Chrf  | Validation Loss | Wer      |
|:-------------:|:------:|:----:|:-----:|:-----:|:---------------:|:--------:|
| 1.9416        | 0.2155 | 100  | 13.09 | 26.48 | 1.7899          | 104.4575 |
| 1.5186        | 0.4310 | 200  | 18.6  | 35.75 | 1.5696          | 87.5732  |
| 1.2884        | 0.6466 | 300  | 17.57 | 37.19 | 1.4751          | 87.2580  |
| 1.0729        | 0.8621 | 400  | 17.92 | 38.23 | 1.4345          | 99.2346  |
| 0.4574        | 1.0776 | 500  | 22.48 | 39.17 | 1.5585          | 83.1607  |
| 0.4517        | 1.2931 | 600  | 22.53 | 38.38 | 1.5763          | 81.7650  |
| 0.4385        | 1.5086 | 700  | 20.05 | 39.46 | 1.5852          | 96.8483  |
| 0.3934        | 1.7241 | 800  | 26.89 | 42.67 | 1.5332          | 70.6889  |
| 0.3587        | 1.9397 | 900  | 28.95 | 44.16 | 1.5025          | 64.9707  |
| 0.1528        | 2.1552 | 1000 | 28.32 | 42.36 | 1.5882          | 65.8712  |
| 0.1425        | 2.3707 | 1100 | 25.5  | 42.42 | 1.6056          | 75.0113  |
| 0.1389        | 2.5862 | 1200 | 26.52 | 42.11 | 1.6236          | 70.6439  |
| 0.1532        | 2.8017 | 1300 | 25.78 | 41.61 | 1.6196          | 75.9118  |
| 0.1138        | 3.0172 | 1400 | 26.01 | 40.88 | 1.7185          | 69.6983  |
| 0.0661        | 3.2328 | 1500 | 28.74 | 43.16 | 1.6626          | 71.2292  |
| 0.0625        | 3.4483 | 1600 | 29.16 | 43.6  | 1.6835          | 66.3215  |
| 0.0615        | 3.6638 | 1700 | 28.93 | 44.08 | 1.6756          | 68.3476  |
| 0.0611        | 3.8793 | 1800 | 27.77 | 43.67 | 1.6648          | 72.1747  |
| 0.0344        | 4.0948 | 1900 | 28.33 | 44.18 | 1.7351          | 68.1225  |
| 0.0339        | 4.3103 | 2000 | 28.9  | 42.98 | 1.7715          | 67.0869  |
| 0.0369        | 4.5259 | 2100 | 29.83 | 44.87 | 1.7200          | 64.8807  |
| 0.0326        | 4.7414 | 2200 | 28.23 | 43.75 | 1.7232          | 69.3832  |
| 0.0346        | 4.9569 | 2300 | 27.72 | 43.1  | 1.7688          | 72.8050  |
| 0.0167        | 5.1724 | 2400 | 28.73 | 43.26 | 1.8072          | 67.4471  |
| 0.0146        | 5.3879 | 2500 | 29.91 | 44.24 | 1.7801          | 66.4566  |
| 0.0165        | 5.6034 | 2600 | 29.34 | 44.33 | 1.7782          | 68.2125  |
| 0.0143        | 5.8190 | 2700 | 27.78 | 43.07 | 1.7675          | 72.5799  |
| 0.0106        | 6.0345 | 2800 | 29.45 | 43.31 | 1.7660          | 67.5371  |
| 0.0098        | 6.25   | 2900 | 27.89 | 42.67 | 1.7803          | 71.6344  |
| 0.0087        | 6.4655 | 3000 | 27.66 | 43.04 | 1.7786          | 72.0396  |
| 0.0089        | 6.6810 | 3100 | 1.7661| 29.81 | 44.65           | 67.3120  |
| 0.0081        | 6.8966 | 3200 | 1.7744| 29.48 | 44.3            | 68.0324  |
| 0.0095        | 7.1121 | 3300 | 1.8197| 29.55 | 44.2            | 67.5371  |
| 0.0112        | 7.3276 | 3400 | 1.8102| 29.34 | 43.9            | 66.2765  |
| 0.0075        | 7.5431 | 3500 | 1.8004| 29.57 | 44.43           | 67.3570  |
| 0.0111        | 7.7586 | 3600 | 1.8015| 29.56 | 44.57           | 66.4566  |
| 0.009         | 7.9741 | 3700 | 1.8001| 29.7  | 45.24           | 66.6817  |
| 0.005         | 8.1897 | 3800 | 1.8184| 29.21 | 44.4            | 67.4471  |
| 0.0055        | 8.4052 | 3900 | 1.8222| 29.67 | 44.35           | 67.1319  |
| 0.0042        | 8.6207 | 4000 | 1.8231| 29.51 | 44.29           | 67.0869  |


### Framework versions

- Transformers 4.40.2
- Pytorch 2.2.0+cu121
- Datasets 2.19.1
- Tokenizers 0.19.1