Thank you unsloth for the amazing work once again! The new sleep mode in vLLM is really amazing, and it seems the community hasn’t quite wrapped their heads ar

Thank you unsloth for the amazing work once again! The new sleep mode in vLLM is really amazing, and it seems the community hasn’t quite wrapped their heads around how much more accessible this makes RL training. I’m reading a lot of dismaying posts on this thread, pushing the idea that only big labs should be doing RL. This couldn’t be further from the truth, folks should try it for themselves and see the outcomes!