rahulunair

Personal SGLang experiments and model-specific recipes for Arc Pro B60, B65, and B70 GPUs.

vLLM-Omni on Intel XPU with native Z-Image CFG-parallel validation

Intel XPU SGLang LLM and diffusion images

Intel XPU vLLM image validated with TP=2

Ollama on Intel GPUs via the llama.cpp SYCL/oneAPI backend. Verified on Arc Pro B70.

Optimized llama.cpp server for Intel GPUs (Arc, Arc Pro, Data Center) with SYCL

Intel LLM Scaler vLLM - MoE/FP8 Fix for Intel XPU

Pytorch and transformers for intel discrete GPU.

Tensorflow with VNNI patches on Clear Linux

Vanilla Tensorflow 1.12 on ClearLinux 26270