BigHugger
GH Repository · TokenRhythm

NeoHorse

NeoHorse-1: Towards Recursive Self-Improvement via Agentic Post-Training with Routing Harness.

stars
486
30-day movement
+18762/day
Related entries
60
Connections
1
recursive-self-improvementpost-trainingllmagent-apprsilarge-language-modelsagent

NeoHorse is a GitHub repository by TokenRhythm described as 'NeoHorse-1: Towards Recursive Self-Improvement via Agentic Post-Training with Routing Harness.' Based on its name and topics, it appears to be an agent application exploring recursive self-improvement (RSI) of large language models through post-training guided by a routing harness.

Reach for it if you want to study or experiment with an open (Apache-2.0) implementation of agentic post-training aimed at recursive self-improvement.

Use it to

  • Explore agentic post-training techniques
  • Study recursive self-improvement approaches
  • Examine the routing harness design
  • Fork and build on an Apache-2.0 agent codebase

For Researchers and developers exploring LLM self-improvement and post-training

Role
agent-app
Licence
Apache-2.0
Forks
8
Last push
2026-09-16
topicsagentllmpost-trainingrecursive-self-improvementrsi