FoPO trains LLMs for strategic reasoning by combining self-interest with opponent modeling in policy optimization, yielding gains on two new datasets and better out-of-domain generalization than standard baselines.
late.” Kindly refine your consideration to those objects which simultaneously exhibit both “loud
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Foresight Optimization for Strategic Reasoning in Large Language Models
FoPO trains LLMs for strategic reasoning by combining self-interest with opponent modeling in policy optimization, yielding gains on two new datasets and better out-of-domain generalization than standard baselines.