Paper Summary: This post introduces Murasaki, an open-source CoT translation model tailored for the ACGN domain, complete with an out-of-the-box local inference GUI.
Summary Of Strengths: The project demonstrates exceptional practical utility.
Summary Of Weaknesses: Where is the technical report?
Comments Suggestions And Typos: Have you considered or experimented with Reinforcement Learning (RL) post-training methods, such as GRPO or PPO?
Tohrusky 最近的时间轴更新
Tohrusky
Select Language