A training-time LLM that refines actions and reshapes rewards, co-trained with an RL controller through a shared buffer, reports large gains over RL and LLM baselines on three L2RPN power grid challenges.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Think Twice, Act Once: A Co-Evolution Framework of LLM and RL for Large-Scale Decision Making
A training-time LLM that refines actions and reshapes rewards, co-trained with an RL controller through a shared buffer, reports large gains over RL and LLM baselines on three L2RPN power grid challenges.