Counterexample-guided refinement and agentic loops raise LLM regex induction success from 3-39% to 38-74% on difficult task groups.
Pranav Garg, Christof Löding, P
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
-
Counterexample Guided Learning in the Large using Reasoning Agents
Counterexample-guided refinement and agentic loops raise LLM regex induction success from 3-39% to 38-74% on difficult task groups.