ReLoop closes the feasibility-correctness gap in LLM optimization code via structured generation and behavioral verification with parameter perturbations, reaching 100% executability and accuracy gains on benchmarks while releasing RetailOpt-190.
## Data Structure / Thedata variable is PRE-DEFINED with these keys: {schema} /CRITICAL: Do NOT create data = {...}. Just use data[
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SE 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
ReLoop: Structured Modeling and Behavioral Verification for Reliable LLM-Based Optimization
ReLoop closes the feasibility-correctness gap in LLM optimization code via structured generation and behavioral verification with parameter perturbations, reaching 100% executability and accuracy gains on benchmarks while releasing RetailOpt-190.