---
license: apache-2.0
base_model: openai/whisper-large
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: whisper-large-children-ft
  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-children-ft

This model is a fine-tuned version of [openai/whisper-large](https://huggingface.co/openai/whisper-large) on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 0.8308
- Wer: 0.7942
- Cer: 0.6731

## 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: 16
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- training_steps: 5000
- mixed_precision_training: Native AMP

### Training results

| Training Loss | Epoch | Step | Validation Loss | Wer    | Cer    |
|:-------------:|:-----:|:----:|:---------------:|:------:|:------:|
| 0.9663        | 0.2   | 1000 | 0.9738          | 1.1973 | 1.0267 |
| 0.8767        | 0.4   | 2000 | 0.9192          | 0.9408 | 0.8104 |
| 0.8277        | 0.59  | 3000 | 0.8746          | 0.8052 | 0.6862 |
| 0.6799        | 0.79  | 4000 | 0.8483          | 0.8404 | 0.7421 |
| 0.6919        | 0.99  | 5000 | 0.8308          | 0.7942 | 0.6731 |


### Framework versions

- Transformers 4.39.3
- Pytorch 2.1.1+cu118
- Datasets 3.5.1
- Tokenizers 0.15.2