Magnus Müller 4a4336e1a1 Improve judge system evaluation criteria ...

- Remove verbose architecture documentation comments
- Add guideline to avoid file system usage for short tasks
- Add stricter evaluation criteria for blocked/impossible tasks
- Add criteria for bad trajectories and tool usage
- Add requirement for task completion scoring
- Add trajectory optimization considerations

2025-06-26 15:01:10 +02:00

..

judge_system.py

Improve judge system evaluation criteria

2025-06-26 15:01:10 +02:00

service.py

Complete last_message flow to comprehensive judge

2025-06-25 10:34:50 +02:00