Environment setup

[zackzhou0814]

Hi,

If possible, could you please provide the instructions or the exact software versions required to build the environment? There are many potential conflicts, and having the correct versions would help prevent problems. Thank you very much!

[Chesten223]

I just finished installing and validating Evo2 on my 5070 Ti mobile GPU. Here are my specs:
CUDA 13.0 cuDNN 9.13.0

pip3 install --pre torch torchvision --index-url https://download.pytorch.org/whl/nightly/cu130

pip install transformer_engine[pytorch]==2.5.0

pip install flash-attn --no-build-isolation -v

Hope this helps!

[Chesten223]

USE THIS !!!
curl -LsSf https://astral.sh/uv/install.sh | INSTALLER_DOWNLOAD_URL=https://wheelnext.astral.sh sh
source $HOME/.local/bin/env
uv pip install evo2
uv pip install transformer-engine[pytorch]
uv pip install psutil
uv pip install flash_attn --no-build-isolation

It's VERY EASY! Wheel Variants NB!

[akins-dev]

THIS WORKED FOR ME

Upgrade pip, setuptools, wheel

pip install --upgrade pip setuptools wheel

Install Torch (must match CUDA 12.4 runtime)

pip install torch==2.8.0
pip install flash-attn --no-build-isolation
pip uninstall -y transformer-engine transformer_engine
pip install 'transformer_engine[pytorch]>=1.13' --no-build-isolation

[ddofer]

An y luck getting this to install on aarm64 (Nvidia dgx)?