metadata
library_name: peft
language:
- nl
license: apache-2.0
base_model: openai/whisper-large-v2
tags:
- base_model:adapter:openai/whisper-large-v2
- lora
- transformers
datasets:
- jasmin
- jasmin-cgn
metrics:
- wer
model-index:
- name: whisper-new-nnat-20h-meta-seed2025
results:
- task:
type: automatic-speech-recognition
name: Automatic Speech Recognition
dataset:
name: JASMIN-CGN
type: jasmin
metrics:
- type: wer
value: 20.039338734604037
name: Wer
whisper-new-nnat-20h-meta-seed2025
This model is a fine-tuned version of openai/whisper-large-v2 on the JASMIN-CGN dataset. It achieves the following results on the evaluation set:
- Loss: 0.4638
- Wer: 20.0393
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: 48
- eval_batch_size: 32
- seed: 42
- optimizer: Use adamw_torch_fused with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 39
- num_epochs: 3.0
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 1.2511 | 0.3 | 39 | 0.9054 | 41.8161 |
| 0.6972 | 0.6 | 78 | 0.5555 | 23.7203 |
| 0.5503 | 0.9 | 117 | 0.5132 | 23.7765 |
| 0.4966 | 1.2 | 156 | 0.4943 | 21.6504 |
| 0.4841 | 1.5 | 195 | 0.4807 | 20.8027 |
| 0.4759 | 1.8 | 234 | 0.4732 | 20.3625 |
| 0.4804 | 2.1 | 273 | 0.4682 | 20.3531 |
| 0.4416 | 2.4 | 312 | 0.4672 | 21.1071 |
| 0.4385 | 2.7 | 351 | 0.4652 | 20.1798 |
| 0.4327 | 3.0 | 390 | 0.4638 | 20.0393 |
Framework versions
- PEFT 0.17.1
- Transformers 4.57.6
- Pytorch 2.8.0+cu128
- Datasets 4.5.0
- Tokenizers 0.22.2