hugging-face-cli-with-codespaces

Repo that allows me to build AI tools on top of Hugging Face and OpenAI Whisper

Try out fine-tuning

Run this script

• Fine-tuning HuggingFace Tutorial

"When you use a pretrained model, you train it on a dataset specific to your task. This is known as fine-tuning, an incredibly powerful training technique. In this tutorial, you will fine-tune a pretrained model with a deep learning framework of your choice:"

Tried conda with and failed:

https://huggingface.co/docs/transformers/training

How to do fast inference using API

• Use hugging face inference api
• How to use inference

Monitor GPU while Training

Use split terminal and do:

nvidia-smi -l 1 in one htop in another

Verify GPU working

• Reference PyTorch site
• numba -s | grep cuda
• run python utils/verify_cuda_pytorch.py
• run nvidia-smi should show a GPU

Notes on OS X M1

conda create -n hf python=3.9
conda activate hf
conda env config vars set CONDA_SUBDIR=osx-arm64
#install nightly
pip install -U --pre torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/nightly/cpu
#install rust if not installed
curl — proto ‘=https’ — tlsv1.2 -sSf https://sh.rustup.rs | sh
#install via brew
brew install cmake
brew install pkg-config

References

• Monitoring GPU Training