Search papers, labs, and topics across Lattice.
This paper introduces J-Zero, a unified framework for Challenger-Solver-Judge co-evolution that enables self-improvement of language models in both verifiable and unverifiable domains without requiring human supervision. The method features an adversarial interaction where the Challenger generates complex tasks and the Solver enhances its responses, while the Judge co-adapts using known preference pairs rather than subjective scoring. J-Zero demonstrates significant performance improvements, outperforming baselines by an average of 4.2 points in verifiable domains and 8.0 points in unverifiable domains, with sustained improvement over multiple iterations.
J-Zero achieves remarkable self-improvement in language models, outperforming traditional methods by leveraging zero data for Judge co-adaptation.
Self-evolving language models have recently emerged as a promising path toward superintelligence, with the advantage of reducing the cost of human supervision. While considerable progress has been made in verifiable domains, self-evolution in unverifiable domains remains substantially less explored. We propose Judge co-adaptation from Zero data (J-Zero), a unified Challenger--Solver--Judge co-evolution framework that supports self-improvement across both domains. The Challenger and Solver co-evolve through an adversarial interaction: the Challenger generates increasingly difficult tasks, while the Solver learns to produce higher-quality responses to them. In parallel, the Judge co-adapts using preference pairs whose ordering is known in advance from how each response was produced, i.e., the Solver's answer over the Challenger's, and its decomposed-and-recombined answer over its one-shot answer, rather than from the Judge's own scores. J-Zero outperforms the baselines by an average of 4.2 points on verifiable and 8.0 points on unverifiable domains, and continues to improve through at least ten iterations, whereas the baselines degrade after two.
