Extended best-of-N sampling with Tsallis statistics allows flexible empowerment in RL reasoning, balancing exploration-exploitation and improving locomotion task performance.
Its gain is designed asα(1−γ)
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
-
Flexible Empowerment at Reasoning with Extended Best-of-N Sampling
Extended best-of-N sampling with Tsallis statistics allows flexible empowerment in RL reasoning, balancing exploration-exploitation and improving locomotion task performance.