Coherent IRL learns dense rewards from demos to enable sample-efficient off-policy improvement of large behavior-cloned policies on sparse robotic manipulation tasks.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Coherent Off-Policy Improvement of Large Behavior Models with Learned Rewards
Coherent IRL learns dense rewards from demos to enable sample-efficient off-policy improvement of large behavior-cloned policies on sparse robotic manipulation tasks.