metadata
language:
- sr
license: apache-2.0
tags:
- whisper-event
- generated_from_trainer
datasets:
- mozilla-foundation/common_voice_11_0,google/fleurs
metrics:
- wer
model-index:
- name: Whisper medium Serbian El Greco
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: mozilla-foundation/common_voice_11_0,google/fleurs sr,sr_rs
type: mozilla-foundation/common_voice_11_0,google/fleurs
config: sr
split: None
metrics:
- name: Wer
type: wer
value: 12.140833670578713
Whisper medium Serbian El Greco
This model is a fine-tuned version of openai/whisper-medium on the mozilla-foundation/common_voice_11_0,google/fleurs sr,sr_rs dataset. It achieves the following results on the evaluation set:
- Loss: 0.4868
- Wer: 12.1408
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: 3e-06
- train_batch_size: 32
- eval_batch_size: 16
- seed: 42
- distributed_type: multi-GPU
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- training_steps: 10000
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.0222 | 2.72 | 1000 | 0.3442 | 14.0834 |
| 0.0032 | 5.43 | 2000 | 0.4106 | 14.5285 |
| 0.0011 | 8.15 | 3000 | 0.4331 | 12.8693 |
| 0.0029 | 10.87 | 4000 | 0.3948 | 12.6265 |
| 0.0012 | 13.59 | 5000 | 0.4512 | 12.6669 |
| 0.0009 | 16.3 | 6000 | 0.4890 | 12.7479 |
| 0.001 | 19.02 | 7000 | 0.4868 | 12.1408 |
| 0.0016 | 21.74 | 8000 | 0.4780 | 12.7074 |
| 0.0002 | 24.46 | 9000 | 0.4902 | 12.2218 |
| 0.0012 | 27.17 | 10000 | 0.5059 | 12.6669 |
Framework versions
- Transformers 4.26.0.dev0
- Pytorch 2.0.0.dev20221216+cu116
- Datasets 2.7.1.dev0
- Tokenizers 0.13.2