看来坛子里不能指出逻辑漏洞啊您挺难伺候的。
“The release of DeepSeek-R1 is an amazing boon for the community, but they didn’t release everything—although the model weights are open, the datasets and code used to train the model are not .” huggingface上有Open-R1项目,要深挖没有公开的部分这是人家的核心技术。