Automatic Speech Recognition
Transformers
PyTorch
English
whisper
nyansapo_ai-asr-leaderboard
Generated from Trainer
Eval Results (legacy)
# Load model directly
from transformers import AutoProcessor, AutoModelForSpeechSeq2Seq
processor = AutoProcessor.from_pretrained("eai6/whisper-tiny")
model = AutoModelForSpeechSeq2Seq.from_pretrained("eai6/whisper-tiny")Quick Links
whisper-base.en
This model is a fine-tuned version of openai/whisper-tiny on the Azure-dataset dataset. It achieves the following results on the evaluation set:
- Loss: 0.0237
- Wer: 8.5859
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: 2500
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.1945 | 3.11 | 500 | 0.0626 | 18.0808 |
| 0.0627 | 6.21 | 1000 | 0.0292 | 10.5051 |
| 0.0419 | 9.32 | 1500 | 0.0242 | 9.0909 |
| 0.0419 | 12.42 | 2000 | 0.0242 | 8.8889 |
| 0.0502 | 15.53 | 2500 | 0.0237 | 8.5859 |
Framework versions
- Transformers 4.33.0.dev0
- Pytorch 2.0.1
- Datasets 2.14.4
- Tokenizers 0.13.3
- Downloads last month
- 6
Model tree for eai6/whisper-tiny
Base model
openai/whisper-tinyEvaluation results
- Wer on Azure-datasettest set self-reported8.586
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("automatic-speech-recognition", model="eai6/whisper-tiny")