Mahal Systems TutorBench
Fits high-school and AP-level tutoring assistants. Borrow the per-sample rubric pattern, where each prompt ships with its own expert-written checklist that an LLM judge ticks off.
What it asks the agent to do
- explain a concept adapted to a specific student confusion
- give actionable feedback on a student's work
- write a hint that moves the student forward without solving it
What to borrow from it
grading designcase template
Caveats
Single-turn only; grading is entirely LLM-judge against rubrics; prompts and student states are expert-written, not real students; dataset is on Hugging Face but no runner code or license was found.
Details
| Workloads | education, knowledge qa |
|---|---|
| Interaction | single turn |
| Environment | static |
| End state checked | No |
| Tools | none |
| Grading | llm judge |
| Size | 1,490 |
| Data | synthetic |
| Personal data | none |
| License | unknown |
| Harness | none |
| Status | active |