SWE-Trail trajectory collection method improves coding agent training effectiveness
The story so far
qwen's SWE-Trail method improves the trajectory collection effect of the coding agent's training by utilizing past execution trajectories to guide subsequent exploration, without exposing historical information to the coding agent. This method, using a student model with 124B parameters, increased the average problem-solving rate by 2.33 and 1.60 percentage points on SWE-bench Multilingual and SWE-bench Pro, respectively.
AI-written from the reports · Updated 15 days ago
Coverage timeline
Follow the coverage to see every side of the story.
- Hugging Face Daily PapersSWE-Trail: History-Guided Trajectory Collection for Coding Agent Training
SWE-Trail leverages past execution trajectories to guide subsequent exploration without exposing historical information to coding agents, thereby enhancing the trajectory collection effectiveness for coding agent training. This method improves the average problem-solving rate by 2.33 and 1.60 percentage points on SWE-bench Multilingual and SWE-bench Pro, respectively, using a student model with 124B parameters.