Our paper “Confidence-Orchestrated Self-Evolution against Uncertain LLM Feedback” was accepted to EMNLP 2026.