Fix disaggregated prefill/decode RayService for H200 GPU clusters#28

Open

iankouls-aws wants to merge 3 commits into

mainfrom

disaggregated-inference

Commits on Mar 31, 2026

Add disaggregated prefill/decode inference example
iankouls-aws
committed

Commits on Apr 1, 2026

fix(rayservice): use Qwen2.5-7B-Instruct, H200 accelerator, disable head GPUs
iankouls-aws
committed
fix(rayservice): use ray-llm:latest-py311-cu128 image tag
iankouls-aws
committed