Нет описания

Lengyue afc854ad85 Clean unused code & reorganize 2 лет назад
data_server 57755e1e74 Add rust data server 2 лет назад
fish_speech afc854ad85 Clean unused code & reorganize 2 лет назад
tools afc854ad85 Clean unused code & reorganize 2 лет назад
.dockerignore 57755e1e74 Add rust data server 2 лет назад
.gitignore 57755e1e74 Add rust data server 2 лет назад
.pre-commit-config.yaml f1ea577c4c [pre-commit.ci] pre-commit autoupdate 2 лет назад
LICENSE 62710e34a4 Add llama inference tool chain 2 лет назад
README.md 57755e1e74 Add rust data server 2 лет назад
dockerfile f8cfbb4ac0 Greatly improved data loading speed 2 лет назад
pyproject.toml 57755e1e74 Add rust data server 2 лет назад
pyrightconfig.json f7f2c03282 Support pytorch lightning 2 лет назад

README.md

Fish Speech

This repo is still under construction. Please check back later.

Setup

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

# Install flash-attn (for linux)
pip3 install ninja && MAX_JOBS=4 pip3 install flash-attn --no-build-isolation

# Install fish-speech
pip3 install -e .

Rust Data Server

Since loading and shuffle the dataset is very slow and memory consuming, we use a rust server to load and shuffle the dataset. The server is based on GRPC and can be installed by

cd data_server
cargo build --release

Credits