其中long-CoT版本在关键评测基准上足够和o1抗衡，有在Code和Math方面甚至超过了o1，注意，这可是满血o1，不是o1-mini。 而short-CoT版本，对标的是GPT-4o ...

Carnegie Mellon University researchers propose a new LLM training technique that gives developers more control over chain-of-thought length.

Companies can freely deploy Light-R1-32B in commercial products, maintaining full control over their innovations.