Allow uneven shards for multi-GPU inference in vllm backend#494

SahilJain314 merged 2 commits into

mainNVIDIA-NeMo/RL:mainfrom

zhaochengz/multi-gpu_evalNVIDIA-NeMo/RL:zhaochengz/multi-gpu_evalCopy head branch name to clipboard