shunzou05/TerminalHorizon-3K

TerminalHorizon

Scaling Agentic Data for Long-Horizon Terminal Intelligence

Paper · Citation

TerminalHorizon is a fully automated data synthesis engine for long-horizon terminal agents. Grounded in real-world professional work, it constructs executable environments and generates training trajectories that connect agentic behavioral patterns across stages toward a shared goal.

This repository releases the 2,852 training trajectories underlying the project, provided as multi-turn conversations for supervised fine-tuning. The companion TerminalHorizon-Environment repository provides the executable task environments.

Resources

Resource	Description
TerminalHorizon-Environment	Executable terminal-task environments.
TerminalHorizon-3K	Long-horizon training trajectories.
TerminalHorizon-27B	Fine-tuned Qwen3.5-27B.
TerminalHorizon-35B-A3B	Fine-tuned Qwen3.5-35B-A3B.
TerminalHorizon-122B-A10B	Fine-tuned Qwen3.5-122B-A10B.

Dataset Overview

Property	Value
Training trajectories	2,852
Task environment collection	1,535
Professional domains	25
Subdomains	463

The terminal is a shared interaction interface, not a restriction to software tasks. The environments span software engineering, machine learning, education, agriculture, healthcare, finance, scientific research, and other professional domains. They are designed to support substantial, sustained terminal work whose difficulty follows from the professional objective and the work needed to complete it.

TerminalHorizon uses challenge-guided reconstruction and progressive horizon expansion to develop environments for long-horizon problem solving. Its focus is on meaningful work and connected agent behavior, rather than task length alone.

Trajectories are collected with DeepSeek-V4-Pro-0813 using Terminus2, then filtered for training. We also conduct benchmark decontamination before training.

Load the Dataset

from datasets import load_dataset

dataset = load_dataset(
    "shunzou05/TerminalHorizon-3K",
    split="train",
)
print(len(dataset))  # 2852
print(dataset[0]["conversations"])

Citation

@misc{zou2026scalingagenticdatalonghorizon,
      title={Scaling Agentic Data for Long-Horizon Terminal Intelligence},
      author={Shun Zou and Ziyu Ma and Yi Zou and Yong Wang and Lin Chen and Zehui Chen and Guanhua Chen and Xiangxiang Chu and Feng Zhao},
      year={2026},
      eprint={2610.13166},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2610.13166},
}