Aucune description

Lengyue 01cf9bbafd Fix dataset and training loop il y a 2 ans
preparing_data 9adf687fb1 Update base training code il y a 2 ans
speech_lm 01cf9bbafd Fix dataset and training loop il y a 2 ans
.dockerignore 70797ef7d5 Add dockerignore il y a 2 ans
.gitignore 7650b2ac43 Update much robust dataloader il y a 2 ans
.pre-commit-config.yaml e4ce816175 Update deps il y a 2 ans
.project-root aa0a6be778 init training code il y a 2 ans
README.md e4ce816175 Update deps il y a 2 ans
dockerfile 7650b2ac43 Update much robust dataloader il y a 2 ans
ds_config.json db0aa1a99a Init preprocessing tools il y a 2 ans
requirements.txt 7650b2ac43 Update much robust dataloader il y a 2 ans
setup.py 7650b2ac43 Update much robust dataloader il y a 2 ans
train.sh a0a5c3e95c Update training script il y a 2 ans

README.md

Speech LLM

Setup

# Basic environment setup
conda create -n speech-llm python=3.10
conda activate speech-llm
conda install pytorch torchvision torchaudio pytorch-cuda=12.1 -c pytorch -c nvidia

# Install requirements
pip3 install -r requirements.txt

# Install flash-attn
MAX_JOBS=4 pip install flash-attn --no-build-isolation