In this session we will generate images using stable diffusion via huggingface and OpenAI's Dall-E.
Before installing anything, it is recommended to create a new environment
mamba create --name tea4 python=3.9 devbio-napari pyqt -c conda-forge
and activating it:
mamba activate tea4
In order to use Dall-E, we need to install openai:
pip install openai==1.30.1
To test the huggingface models, we need to install those packages:
pip install diffusers==0.24.0 transformers accelerate
For running Huggingface models on an NVidia-compatible GPU, you need to install these libraries:
mamba install pytorch torchvision torchaudio pytorch-cuda=11.8 cudatoolkit=11.8 huggingface_hub -c pytorch -c nvidia -c conda-forge
In case of installation or runtime issues, compare your environment using conda list with the environment stored in the environment.yml file and focus on the libraries which are mentioned in the error messages.