Was able to make an LLM work on integrated Intel GPU (Core Ultra 7 155H) on Fedora. Got about 25% improvement compared to running the model on a CPU.
I'll share more in an upcoming blog post.
#ai #llm #intel #linux