NeoHorse-1：通过带路由 Harness 的智能体后训练迈向递归自我改进
NeoHorse-1: Towards Recursive Self-Improvement via Agentic Post-Training with Routing Harness
NeoHorse-1 是面向递归自我改进的智能体原生模型系列，通过异构模型池与智能路由，将每轮交互记录转化为保留推理、工具调用与 harness 上下文的训练样本。
Authors:NeoHorse Team, Guoliang Cao, Guohao Dai, Tianyu Guo, Kai Han, Hailin Hu, Zihan Jiang, Xiang Kuang, Boxun Li, Yulong Li, Zehua Pei, Yuchuan Tian, Jiamin Wang, Yu Wang, Yunhe Wang, Yihong Wu, Haiyang Xu, Shuo Zhang, Hang Zhou, Siyang Cheng, Jiayu Fan, Wei He, Qingrui Jiao, Hongguang Li, Zhiyuan Li, Runke Liu, Xi Liu, Xinchen Liu, Sinno Jialin Pan, Yi Ren, Liuyang Song, Chenyu Wang, Bei Yu, Quanlu Zhang, Xiangyu Zhang, Mengyu Zheng, Yingjie Zong
Abstract:Recursive self-improvement (RSI) requires a concrete mechanism through which an AI system observes its capabilities and converts that evidence into the next round of learning. We present NeoHorse-1, a family of agent-native models developed to explore this path through agentic post-training. Our system combines a heterogeneous model pool with intelligent routing, recording the predicted capability demand, selected service tier, and subsequent interaction for each user turn. These records are converted into training examples that preserve interleaved reasoning, tool calls, and harness context, and are admitted through structural validation, six-dimensional semantic evaluation, and subscene-level labeling. Routing signals organize supervised fine-tuning into a three-stage curriculum and extend to routing-guided on-policy distillation, where a teacher supervises student-generated responses under the same progression. Capability-guided allocation then converts evaluation feedback into the next training mixture, closing an evaluation-selection-update loop in which what the system learns to do shapes what it learns from next. Across eleven benchmarks covering harness-based agents, tool use, coding, and instruction following, post-training raises the macro-average from 58.94 to 64.87 at 4B and from 65.60 to 69.04 at 9B, substantially narrowing the aggregate gap between the post-trained 4B model and the 9B base model. We further adapt NeoHorse-1-4B into NeoHorse-Jev-4B, a prefill-only model that returns structured decisions and probabilities for agent workflows. NeoHorse-1 provides an initial prototype of this feedback-driven process and a path toward harness-mediated RSI across successive iterations.
| Comments: | Huggingface: this https URL ; Github: this https URL |
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2609.08183 [cs.CL] |
| (or arXiv:2609.08183v2 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2609.08183 arXiv-issued DOI via DataCite |
Submission history
From: Zehua Pei [view email]
[v1]
Tue, 8 Sep 2026 03:14:54 UTC (708 KB)
[v2]
Wed, 14 Oct 2026 13:51:20 UTC (711 KB)
来源：arXiv cs.CL · arxiv.org