---
library_name: transformers
license: apache-2.0
base_model: openai/whisper-large-v3
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: whisper-large-v3-chichewa-variant-b-normalized-transcript
  results: []
---

<!-- 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-large-v3-chichewa-variant-b-normalized-transcript

This model is a fine-tuned version of [openai/whisper-large-v3](https://huggingface.co/openai/whisper-large-v3) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 1.6772
- Wer: 56.0482
- Cer: 25.0424

## 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: 7.5e-06
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 32
- optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 0.05
- training_steps: 3000

### Training results

| Training Loss | Epoch   | Step | Validation Loss | Wer     | Cer     |
|:-------------:|:-------:|:----:|:---------------:|:-------:|:-------:|
| 6.1867        | 0.9259  | 100  | 1.4766          | 83.4464 | 37.9517 |
| 4.3043        | 1.8519  | 200  | 1.1411          | 75.8189 | 36.3040 |
| 3.2145        | 2.7778  | 300  | 1.0356          | 62.7866 | 29.1714 |
| 2.3829        | 3.7037  | 400  | 1.0172          | 61.5115 | 28.0164 |
| 1.7284        | 4.6296  | 500  | 1.0397          | 59.7801 | 27.8698 |
| 1.2327        | 5.5556  | 600  | 1.0958          | 60.7745 | 28.9654 |
| 0.7728        | 6.4815  | 700  | 1.1447          | 58.3645 | 26.6423 |
| 0.5727        | 7.4074  | 800  | 1.1959          | 57.0192 | 26.0145 |
| 0.3477        | 8.3333  | 900  | 1.2624          | 58.1072 | 26.7461 |
| 0.2305        | 9.2593  | 1000 | 1.3064          | 57.1479 | 25.9058 |
| 0.1460        | 10.1852 | 1100 | 1.3492          | 57.2649 | 25.7987 |
| 0.1251        | 11.1111 | 1200 | 1.4039          | 57.3701 | 25.8580 |
| 0.1346        | 12.0370 | 1300 | 1.4000          | 56.7735 | 25.5235 |
| 0.0917        | 12.9630 | 1400 | 1.4034          | 56.6097 | 25.8201 |
| 0.0915        | 13.8889 | 1500 | 1.4139          | 56.4109 | 25.3736 |
| 0.0489        | 14.8148 | 1600 | 1.4882          | 56.4810 | 25.7773 |
| 0.0301        | 15.7407 | 1700 | 1.5243          | 56.2588 | 25.5400 |
| 0.0265        | 16.6667 | 1800 | 1.5395          | 56.1886 | 25.2665 |
| 0.0177        | 17.5926 | 1900 | 1.5489          | 54.9719 | 24.4625 |
| 0.0142        | 18.5185 | 2000 | 1.5950          | 55.5218 | 24.9469 |
| 0.0073        | 19.4444 | 2100 | 1.6268          | 55.7674 | 25.1298 |
| 0.0065        | 20.3704 | 2200 | 1.6454          | 56.0014 | 24.8892 |
| 0.0052        | 21.2963 | 2300 | 1.6621          | 55.0304 | 24.8035 |
| 0.0031        | 22.2222 | 2400 | 1.6772          | 56.0482 | 25.0424 |


### Framework versions

- Transformers 5.8.1
- Pytorch 2.6.0+cu124
- Datasets 3.6.0
- Tokenizers 0.22.2