نسخة أولية وصول مفتوح
Recursive Harness Self-Improvement for Frontier Reasoning Data Synthesis
Generating progressively harder reasoning problems requires synthesis procedures that adapt as the task distribution evolves. Existing task-level recursion reuses generated problems as seeds but leaves the construction harness unchanged. We present task-harness co-evolution, a framework for recursive harness self-impro …