Large-Scale Terminal Agentic Trajectory Generation from Dockerized Environments
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866915771306737664 |
|---|---|
| author | Wu, Siwei Li, Yizhi Song, Yuyang Zhang, Wei Wang, Yang Batista-Navarro, Riza Yang, Xian Tang, Mingjie Dai, Bryan Yang, Jian Lin, Chenghua |
| author_facet | Wu, Siwei Li, Yizhi Song, Yuyang Zhang, Wei Wang, Yang Batista-Navarro, Riza Yang, Xian Tang, Mingjie Dai, Bryan Yang, Jian Lin, Chenghua |
| contents | Training agentic models for terminal-based tasks critically depends on high-quality terminal trajectories that capture realistic long-horizon interactions across diverse domains. However, constructing such data at scale remains challenging due to two key requirements: \textbf{\emph{Executability}}, since each instance requires a suitable and often distinct Docker environment; and \textbf{\emph{Verifiability}}, because heterogeneous task outputs preclude unified, standardized verification. To address these challenges, we propose \textbf{TerminalTraj}, a scalable pipeline that (i) filters high-quality repositories to construct Dockerized execution environments, (ii) generates Docker-aligned task instances, and (iii) synthesizes agent trajectories with executable validation code. Using TerminalTraj, we curate 32K Docker images and generate 50,733 verified terminal trajectories across eight domains. Models trained on this data with the Qwen2.5-Coder backbone achieve consistent performance improvements on TerminalBench (TB), with gains of up to 20\% on TB~1.0 and 10\% on TB~2.0 over their respective backbones. Notably, \textbf{TerminalTraj-32B} achieves strong performance among models with fewer than 100B parameters, reaching 35.30\% on TB~1.0 and 22.00\% on TB~2.0, and demonstrates improved test-time scaling behavior. All code and data are available at https://github.com/Wusiwei0410/TerminalTraj. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_01244 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Large-Scale Terminal Agentic Trajectory Generation from Dockerized Environments Wu, Siwei Li, Yizhi Song, Yuyang Zhang, Wei Wang, Yang Batista-Navarro, Riza Yang, Xian Tang, Mingjie Dai, Bryan Yang, Jian Lin, Chenghua Computation and Language Training agentic models for terminal-based tasks critically depends on high-quality terminal trajectories that capture realistic long-horizon interactions across diverse domains. However, constructing such data at scale remains challenging due to two key requirements: \textbf{\emph{Executability}}, since each instance requires a suitable and often distinct Docker environment; and \textbf{\emph{Verifiability}}, because heterogeneous task outputs preclude unified, standardized verification. To address these challenges, we propose \textbf{TerminalTraj}, a scalable pipeline that (i) filters high-quality repositories to construct Dockerized execution environments, (ii) generates Docker-aligned task instances, and (iii) synthesizes agent trajectories with executable validation code. Using TerminalTraj, we curate 32K Docker images and generate 50,733 verified terminal trajectories across eight domains. Models trained on this data with the Qwen2.5-Coder backbone achieve consistent performance improvements on TerminalBench (TB), with gains of up to 20\% on TB~1.0 and 10\% on TB~2.0 over their respective backbones. Notably, \textbf{TerminalTraj-32B} achieves strong performance among models with fewer than 100B parameters, reaching 35.30\% on TB~1.0 and 22.00\% on TB~2.0, and demonstrates improved test-time scaling behavior. All code and data are available at https://github.com/Wusiwei0410/TerminalTraj. |
| title | Large-Scale Terminal Agentic Trajectory Generation from Dockerized Environments |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2602.01244 |