shunzou05/TerminalHorizon-27B
TerminalHorizon-27B by shunzou05 is a 27 billion parameter language model, fine-tuned from Qwen3.5-27B, designed for long-horizon terminal agent intelligence. It specializes in generating agentic behavioral patterns across various stages to achieve complex goals, leveraging a dataset of real-world professional terminal tasks. This model excels in automating data synthesis for agents operating in diverse professional domains, including software engineering, machine learning, and finance.
Loading preview...
TerminalHorizon-27B Overview
TerminalHorizon-27B is a 27 billion parameter language model developed by shunzou05, specifically fine-tuned from the Qwen3.5-27B base model. Its core purpose is to enable long-horizon terminal agent intelligence through automated data synthesis. The model is trained on the TerminalHorizon-3K dataset, which comprises 2,852 long-horizon training trajectories derived from real-world professional terminal tasks.
Key Capabilities
- Automated Data Synthesis: Generates training trajectories that connect agentic behavioral patterns across multiple stages towards a shared objective.
- Long-Horizon Task Execution: Designed to handle complex, multi-step tasks within a terminal environment.
- Broad Domain Coverage: Training environments span 25 professional domains and 463 subdomains, including software engineering, machine learning, education, healthcare, and finance.
- Agentic Behavior Modeling: Focuses on meaningful work and connected agent behavior rather than just task length.
Good For
- Developing and training AI agents that operate within terminal environments.
- Applications requiring agents to perform complex, multi-stage tasks in professional settings.
- Research into agentic AI, particularly for long-horizon planning and execution.
- Creating intelligent automation solutions for diverse professional workflows.