Automatic Speech Recognition
Transformers
PyTorch
TensorBoard
Safetensors
Korean
whisper
Generated from Trainer
Eval Results (legacy)
Instructions to use byoussef/whisper-large-v2-Ko with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use byoussef/whisper-large-v2-Ko with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("automatic-speech-recognition", model="byoussef/whisper-large-v2-Ko")# Load model directly from transformers import AutoProcessor, AutoModelForSpeechSeq2Seq processor = AutoProcessor.from_pretrained("byoussef/whisper-large-v2-Ko") model = AutoModelForSpeechSeq2Seq.from_pretrained("byoussef/whisper-large-v2-Ko") - Notebooks
- Google Colab
- Kaggle
H/W resources
#4
by r2d209 - opened
I want to know H/W resources you used to train this model.
like GPU(a100 or... something else), GPU RAM size
Yes, it was trained on 7 A100 80GB GPUs. But it's a bit of an overkill. It was done mainly cuz I was working with a very large custom dataset.
I have been successful in training the same model also on a T4 GPU using DeepSpeed. And you could use an even smaller GPU if you utilize PEFT