Terminal Agentic Trajectory Generation on TerminalBench
45Pass@8TerminalTraj
Evaluation Results
| Method | Links | |
|---|---|---|
| TerminalTrajBackbone=Qwen2.5-Coder-7B2026.02 | 45 | |
| SETABackbone=Qwen3-8B2026.02 | 12.5 | |
| Qwen2.5-Coder-7BBackbone=Qwen2.5-Coder-7B2026.02 | 8.75 | |
| Qwen3-8BBackbone=Qwen3-8B2026.02 | 7.5 |