RevengeBench evaluates LLMs on recovering executable code policies from behavioral traces in games by designing experimental opponents, achieving 34-72% recovery across models.
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
-
RevengeBench: Reverse Engineering Code-Space Policies from Behavioral Experiments
RevengeBench evaluates LLMs on recovering executable code policies from behavioral traces in games by designing experimental opponents, achieving 34-72% recovery across models.