unreviewed draftNo one has checked this end to end. Treat the math, code, and citations as unverified. Source on GitHub.
Topic: Agentic RL or offline RL (pick one). Goes with: lecture L16 or L19. Status: planned. To be built next.
The plan: either: long-horizon credit assignment in agentic RL and the SWE-bench-style reward design; OR: offline RL distributional shift, the BCQ/CQL/IQL family, and the bridge to DPO.