install.txt

# Step 0 (Prerequisite): Install Tensorflow/ Docker, see https://www.tensorflow.org/install/docker for more details.

# Step 1: Install Docker image (+mount volume // adjust to your needs)
# A100-based system
docker run --ipc=host --gpus all -it --rm --name tensorflow_2_14_0_gpu_jupyter -v /flash/projects:/tf/notebooks -p 8890:8888 -p 6010:6010 -p 6011:6011 -p 6012:6012 tensorflow/tensorflow:2.14.0-gpu-jupyter

# H100-based system
docker run --ipc=host --gpus all -it --rm --name tensorflow_2_16_1_gpu_jupyter -v /flash/projects:/tf/notebooks -p 8890:8888 -p 6010:6010 -p 6011:6011 -p 6012:6012 tensorflow/tensorflow:2.16.1-gpu-jupyter

# Step 2: Install Diffusers (0.27.0.dev0) to src // please see File Structure for more details
git clone --branch v0.27.0 https://github.com/huggingface/diffusers.git
cd diffusers
pip install .

# Step 3: Install Dependencies
pip install bitsandbytes
pip install accelerate
pip install transformers==4.37.2
pip install vector-quantize-pytorch
pip install torchac
pip install Ninja
pip install onnxruntime-training
pip install tensorflow-datasets
pip install datasets
pip install torchvision
pip install compressai
pip install torchmetrics
pip install clean-fid
pip install lpips