Adversarial teacher-student RL on computationally equivalent narrative rewrites of coding problems yields modest but consistent pass@1 gains on competitive programming benchmarks.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
AR$^2$: Adversarial Reinforcement Learning for Abstract Reasoning in Large Language Models
Adversarial teacher-student RL on computationally equivalent narrative rewrites of coding problems yields modest but consistent pass@1 gains on competitive programming benchmarks.