mirror of
https://github.com/harivansh-afk/evaluclaude-harness.git
synced 2026-04-19 13:01:46 +00:00
grader, test renderer
This commit is contained in:
parent
9297f0b1ee
commit
e0c36241b0
22 changed files with 1914 additions and 5 deletions
33
prompts/grader-user.md
Normal file
33
prompts/grader-user.md
Normal file
|
|
@ -0,0 +1,33 @@
|
|||
# Grading Request
|
||||
|
||||
## Rubric: {{RUBRIC_NAME}}
|
||||
|
||||
{{RUBRIC_DESCRIPTION}}
|
||||
|
||||
**Passing Threshold:** {{PASSING_THRESHOLD}}%
|
||||
|
||||
### Criteria
|
||||
|
||||
{{CRITERIA_LIST}}
|
||||
|
||||
---
|
||||
|
||||
## Output to Evaluate
|
||||
|
||||
```
|
||||
{{OUTPUT}}
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## Your Task
|
||||
|
||||
1. Evaluate the output against each criterion above
|
||||
2. Provide a score (0.0-1.0) and specific feedback for each
|
||||
3. Calculate the weighted overall score
|
||||
4. Return your assessment as JSON
|
||||
|
||||
Remember:
|
||||
- Cite specific evidence from the output for each score
|
||||
- The overall score must equal the weighted average of criterion scores
|
||||
- Feedback should be actionable and specific
|
||||
Loading…
Add table
Add a link
Reference in a new issue