按年被提到
20252025 · 3 次2026
大模型与语言模型›大模型推理与思维链·arXiv:2508.05004 · cs.LG

R-Zero: Self-Evolving Reasoning LLM from Zero Data

Chengsong Huang, Wenhao Yu, Xiaoyang Wang, Hongming Zhang, Zongxia Li, Ruosen Li, Jiaxin Huang, Haitao Mi, Dong Yu · 2025
开始阅读 ↗PDFalphaXiv 讨论
被提及 · 3 人首次 2025-08-10 · 最近 2025-09-10
3不同的人
2活跃月份
3讨论串
—被引(OpenAlex)
arXiv 摘要(候选层不写导读)

Self-evolving Large Language Models (LLMs) offer a scalable path toward super-intelligence by autonomously generating, refining, and learning from their own experiences. However, existing methods for training such models still rely heavily on vast human-curated tasks and labels, typically via fine-tuning or reinforcement learning, which poses a fundamental bottleneck to advancing AI systems toward capabilities beyond human intelligence. To overcome this limitation, we introduce R-Zero, a fully autonomous framework that generates its own training data from scratch. Starting from a single base LLM, R-Zero initializes two independent models with distinct roles, a Challenger and a Solver. These models are optimized separately and co-evolve through interaction: the Challenger is rewarded for proposing tasks near the edge of the Solver capability, and the Solver is rewarded for solving increasingly challenging tasks posed by the Challenger. This process yields a targeted, self-improving curriculum without any pre-existing tasks and labels. Empirically, R-Zero substantially improves reasoning capability across different backbone LLMs, e.g., boosting the Qwen3-4B-Base by +6.49 on math-reasoning benchmarks and +7.54 on general-domain reasoning benchmarks.

候选层:至少 2 个不同的人、在 2 个不同月份提到过,还没到主书架的 3 人 3 个月。没有做立场标注,所以只写“被提及”。

HN 书评

没有评论直接链到它;下面是它被提交的讨论串里回复最多的评论(不计入统计)
讨论串评论vineethy · 2025-08

“Interesting twist on automated curriculum learning. This paper is using an LLM for the environment and the policy. Other papers use LLMs for policy/value fn. Would be cool to see other reward strategies tying all these threads together”

R-Zero: Self-Evolving Reasoning LLM from Zero Data原评论 ↗
讨论串评论clbrmbr · 2025-09

“Terrible choice of name. DeepSeek developed a historically important model called “R-Zero” (this was the predecessor to R1 that was training without any coldstart SFT, and was very strong but difficult to read chain of thought because it code switches into Chinese and has no line breaks).”

R-Zero: Self-Evolving Reasoning LLM from Zero Data原评论 ↗
讨论串评论Davidzheng · 2025-09

“I think in formal domain like lean it should actually be possible to do it from zero--but seems like no major successes no far”

R-Zero: Self-Evolving Reasoning LLM from Zero Data原评论 ↗
被提交为 story · 3 次