[2603.02218] Self-Play Only Evolves When Self-Synthetic Pipeline Ensures Learnable Information Gain

Computer Science > Machine Learning arXiv:2603.02218 (cs) [Submitted on 10 Feb 2026] Title:Self-Play Only Evolves When Self-Synthetic Pipeline Ensures Learnable Information Gain Authors:Wei Liu, Siya Qi, Yali Du, Yulan He View a PDF of the paper titled Self-Play Only Evolves When Self-Synthetic Pipeline Ensures Learnable Information Gain, by Wei Liu and 3 other authors View PDF HTML (experimental) Abstract:Large language models (LLMs) make it plausible to build systems that improve through self-evolving loops, but many existing proposals are better understood as self-play and often plateau quickly. A central failure mode is that the loop synthesises more data without increasing learnable information for the next iteration. Through experiments on a self-play coding task, we reveal that sustainable self-evolution requires a self-synthesised data pipeline with learnable information that increases across iterations. We identify triadic roles that self-evolving LLMs play: the Proposer, which generates tasks; the Solver, which attempts solutions; and the Verifier, which provides training signals, and we identify three system designs that jointly target learnable information gain from this triadic roles perspective. Asymmetric co-evolution closes a weak-to-strong-to-weak loop across roles. Capacity growth expands parameter and inference-time budgets to match rising learnable information. Proactive information seeking introduces external context and new task sources that prevent s...