研究库 论文知识库

内容库 / 论文

arXiv:2608.26582 · 多模态

J-Zero: Unified Challenger--Solver--Judge Co-Evolution from Zero Data

J-Zero: Unified Challenger--Solver--Judge Co-Evolution from Zero Data

类型：arxiv
标识：2608.26582
链接：https://arxiv.org/abs/2608.26582
主分类：multimodal
形态：method
TLDR：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
待LLM分类：否
来源文件：
/inbox/tom/_candidates/2026-08-31-agent-rag-longcontext-candidates.json

相关 · 按语义匹配

文档 Tom 评 flyP · 2026-07-23 · ABot-World-0 / LongForcing 精读

笔记 LongHorizon-Harness · 短审稿（2026-08-10 15:50 · flyP 精读棒）

笔记 Tom 文献雷达 · Agent × RAG × Long-Context · 2026-06-23（第3次）

文档 Tom-on-flyP · 2026-07-17 互评

仓库 awabahmad469/composite-endpoints-audit

仓库 Zafar-southeast/SPR-SMN

arXiv

2608.26582

TLDR

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

arXiv ↗ 中文全文