Gen AI Playground on Intel® Data Center GPU Max 1100

Welcome to the Gen AI Playground, a collection of notebooks designed to showcase generative AI workloads on Intel® Data Center GPU Max 1100. These notebooks are designed to be accessible to a broad audience, including AI creators, artists, engineers, and those who are just curious about generative AI.

Table of Contents

• Welcome Notebook
• Text-to-Image with Stable Diffusion
• Image-to-Image with Stable Diffusion
• Simple LLM Inference: Playing with Language Models
• Code generation using LLMs
• Gemma model finetuning with SFT and LoRA
• QLoRA finetuning on Intel GPUs
• Please create an issue here if you want me to add new models and implementations.

Requirements

Hardware

• Intel® Data Center GPU Max 1100

Software

• Python
• PyTorch with Intel Extension for PyTorch
• Jupyter Notebook or Jupyter Lab

Installation

Run the following commands to set up the necessary software requirements:

python -m pip install torch==2.0.1a0 torchvision==0.15.2a0 intel_extension_for_pytorch==2.0.110+xpu -f https://developer.intel.com/ipex-whl-stable-xpu

For Jupyter Notebook or Jupyter Lab:

pip install jupyterlab

or

pip install notebook

Each notebook contains a first cell that installs additional packages using conda, which assumes you're running inside a conda environment. Here's a sample installation from one of the notebooks:

import sys
!{sys.executable} -m pip install  invisible-watermark
!conda install -y --quiet -c conda-forge \
    accelerate==0.23.0 \
    validators==0.22.0 \
    diffusers==0.18.2 \
    transformers==4.32.1 \
    tensorboardX \
    pillow \
    ipywidgets \
    ipython \
    sentencepiece

Usage

Simply navigate to one of the notebooks to get started! Each notebook comes with comprehensive guides and examples to help you explore the fascinating world of generative AI.

License

Apache License.See LICENSE for more details.