[fully_async, reward] feat: enable GenRM/DisRM support in fully async training#6044

Open

xiefan46 wants to merge 12 commits intoverl-project:mainfrom

xiefan46:async-genrm

Commits on Apr 17, 2026

feat: enable GenRM/DisRM support in fully async training
xiefan46
committed
revert: restore CLAUDE.md and .gitignore to main state
xiefan46
committed
revert: restore 0.5b training script to main state
xiefan46
committed
simplify: remove unused enable_resource_pool logic and trainer reward loop changes
xiefan46
committed
add assert: disallow use_trainer_do_validate with GenRM/DisRM
xiefan46
committed
refactor: rename _init_reward_loop to _create_reward_loop_manager to avoid overriding base class; add TODO for upstream async fix
xiefan46
committed
test: remove stale TestTrainerInitRewardLoopCondition that no longer matches code
xiefan46
committed