metadata
base_model: openai/whisper-medium
datasets:
- miosipof/asr_en
language:
- en
library_name: peft
license: apache-2.0
metrics:
- wer
tags:
- generated_from_trainer
model-index:
- name: Whisper Medium
results:
- task:
type: automatic-speech-recognition
name: Automatic Speech Recognition
dataset:
name: miosipof/asr_en
type: miosipof/asr_en
config: default
split: train
args: default
metrics:
- type: wer
value: 106.7911714770798
name: Wer
Whisper Medium
This model is a fine-tuned version of openai/whisper-medium on the miosipof/asr_en dataset. It achieves the following results on the evaluation set:
- Loss: 3.9473
- Wer: 106.7912
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: 32
- eval_batch_size: 16
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 128
- training_steps: 1024
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 7.4432 | 4.2667 | 64 | 7.9570 | 167.9117 |
| 6.9871 | 8.5333 | 128 | 7.0858 | 167.5722 |
| 6.1972 | 12.8 | 192 | 6.3333 | 205.2632 |
| 5.9006 | 17.0667 | 256 | 6.0843 | 203.5654 |
| 5.61 | 21.3333 | 320 | 5.8153 | 168.7606 |
| 5.2344 | 25.6 | 384 | 5.4746 | 168.0815 |
| 4.8067 | 29.8667 | 448 | 5.0913 | 168.2513 |
| 4.3927 | 34.1333 | 512 | 4.7586 | 201.5280 |
| 4.1103 | 38.4 | 576 | 4.5158 | 164.5161 |
| 3.8975 | 42.6667 | 640 | 4.3460 | 108.4890 |
| 3.7471 | 46.9333 | 704 | 4.2178 | 109.1681 |
| 3.6146 | 51.2 | 768 | 4.1226 | 108.3192 |
| 3.53 | 55.4667 | 832 | 4.0471 | 107.9796 |
| 3.4579 | 59.7333 | 896 | 3.9927 | 107.4703 |
| 3.4061 | 64.0 | 960 | 3.9594 | 106.9610 |
| 3.3577 | 68.2667 | 1024 | 3.9473 | 106.7912 |
Framework versions
- PEFT 0.12.0
- Transformers 4.44.2
- Pytorch 2.4.1+cu121
- Datasets 3.0.0
- Tokenizers 0.19.1