[Performance] Fix queuing in llm wrappers#3125

vmoens merged 1 commit intomainpytorch/rl:mainfrom

fix-vllm-ultra-batchingpytorch/rl:fix-vllm-ultra-batchingCopy head branch name to clipboard