|
--- |
|
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 |
|
- ymoslem/Tatoeba-Speech-Irish |
|
- ymoslem/Wikimedia-Speech-Irish |
|
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, SpokenWords, Tatoeba, and Wikimedia |
|
type: ymoslem/IWSLT2023-GA-EN |
|
metrics: |
|
- name: Bleu |
|
type: bleu |
|
value: 23.1 |
|
- name: Wer |
|
type: wer |
|
value: 82.89058982440342 |
|
--- |
|
|
|
<!-- 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, SpokenWords, Tatoeba, and Wikimedia dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 1.2172 |
|
- Bleu: 23.1 |
|
- Chrf: 42.54 |
|
- Wer: 82.8906 |
|
|
|
## 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: 32 |
|
- 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: 0.03 |
|
- training_steps: 1000 |
|
- mixed_precision_training: Native AMP |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | Bleu | Chrf | Wer | |
|
|:-------------:|:-----:|:----:|:---------------:|:-----:|:-----:|:--------:| |
|
| 2.8459 | 0.07 | 100 | 2.0769 | 3.28 | 18.43 | 149.0770 | |
|
| 2.3328 | 0.13 | 200 | 1.8396 | 4.5 | 22.06 | 207.7443 | |
|
| 2.1669 | 0.2 | 300 | 1.6215 | 14.6 | 30.8 | 89.1941 | |
|
| 1.8606 | 0.26 | 400 | 1.5030 | 14.65 | 33.33 | 92.4358 | |
|
| 1.7255 | 0.33 | 500 | 1.4085 | 14.9 | 35.14 | 103.8271 | |
|
| 1.5855 | 0.39 | 600 | 1.3587 | 15.78 | 35.02 | 103.0617 | |
|
| 1.5875 | 0.46 | 700 | 1.2986 | 25.3 | 41.37 | 69.4732 | |
|
| 1.44 | 0.53 | 800 | 1.2575 | 25.78 | 42.23 | 70.0585 | |
|
| 1.3317 | 0.59 | 900 | 1.2338 | 23.24 | 41.64 | 79.1085 | |
|
| 1.3166 | 0.66 | 1000 | 1.2172 | 23.1 | 42.54 | 82.8906 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.39.3 |
|
- Pytorch 2.2.1+cu121 |
|
- Datasets 2.18.0 |
|
- Tokenizers 0.15.2 |
|
|
