|
--- |
|
license: apache-2.0 |
|
tags: |
|
- generated_from_trainer |
|
datasets: |
|
- fleurs |
|
metrics: |
|
- wer |
|
model-index: |
|
- name: openai/whisper-large-v2 |
|
results: |
|
- task: |
|
name: Automatic Speech Recognition |
|
type: automatic-speech-recognition |
|
dataset: |
|
name: fleurs |
|
type: fleurs |
|
config: he_il |
|
split: test |
|
args: he_il |
|
metrics: |
|
- name: Wer |
|
type: wer |
|
value: 30.891489669122958 |
|
--- |
|
|
|
<!-- 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. --> |
|
|
|
# openai/whisper-large-v2 |
|
|
|
This model is a fine-tuned version of [openai/whisper-large-v2](https://huggingface.co/openai/whisper-large-v2) on the fleurs dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 0.4945 |
|
- Wer: 30.8915 |
|
|
|
## 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-06 |
|
- train_batch_size: 128 |
|
- eval_batch_size: 32 |
|
- seed: 42 |
|
- gradient_accumulation_steps: 4 |
|
- total_train_batch_size: 512 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: linear |
|
- lr_scheduler_warmup_steps: 10 |
|
- training_steps: 200 |
|
- mixed_precision_training: Native AMP |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | Wer | |
|
|:-------------:|:-----:|:----:|:---------------:|:-------:| |
|
| 0.425 | 24.01 | 50 | 0.4106 | 27.2504 | |
|
| 0.1906 | 49.01 | 100 | 0.4420 | 29.0131 | |
|
| 0.0982 | 74.01 | 150 | 0.4795 | 30.3063 | |
|
| 0.0717 | 99.01 | 200 | 0.4945 | 30.8915 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.26.0.dev0 |
|
- Pytorch 1.13.1+cu117 |
|
- Datasets 2.8.1.dev0 |
|
- Tokenizers 0.13.2 |
|
|
