WIP: polish(pu): add a polished version of qwen prior policy#397

Open

puyuan1996 wants to merge 12 commits intomainfrom

dev-jericho-llm-prior

Commits on Jul 18, 2025

Qwen is tested as a policy in the jericho environment
puyuan
committed

Commits on Jul 24, 2025

fixed the bug that bad reflection cannot be collected
puyuan
committed

Commits on Aug 11, 2025

polish(pu): created a qwen prior policy that can obtain 290 in detective
puyuan
committed
feature (pu): integrated a Qwen prior policy that achieves an average return of 120.5 over 20 episodes, with a maximum return of 300.
puyuan
committed
polish(pu): add a polished args version of qwen prior
puyuan
committed
polish(pu): polish temperature in qwen prior policy
puyuan
committed

Commits on Aug 25, 2025

Commits on Aug 26, 2025

Commits on Aug 27, 2025