原书出处
Build a Reasoning Model (From Scratch)
Exercise 6.1: Adding format-aware reward shaping Extend the reward_rlvr function from this chapter so that it assigns partial credit based on output format. Specifically, modify the reward function so that it returns 1.0 if the model produces the correct answer in the required \boxed{} format, 0.5 if the answer is correct but not boxed, and 0.0 otherwise. 6.7 Preparing learning signals from rollouts via advantages
原书定位详情
来源:a19c7933-4eda-4176-aebd-30fe7e1ad292
正文位置:extracted/Build a Reasoning Model (From Scratch)/_epub-src/OEBPS/Text/chapter-6.html
EPUB SHA-256:52d4d07986020be70f5aac21762dfcf60fcba08db0ba29d4f63957caf12aa377