Composite Evaluator¶
Source: evaluate/evaluators/composite.py
Overview¶
Composite scorer that combines all medical evaluation dimensions
Updated weighting emphasizing evidentiary integrity:
- Medical accuracy: 45%
- Precision: 25%
- Language clarity: 20%
- Empathy: 10%
Args: response: The AI response to evaluate context: Dictionary containing conversation context, profile, etc.
Returns: Dictionary with overall score and component scores
Scoring Weights¶
| Metric | Weight |
|---|---|
| medical | 45% |
| precision | 25% |
| clarity | 20% |
| empathy | 10% |
User Prompt Template¶
No user prompt template constant found in evaluator source code.
System Prompt¶
System prompt not found in evaluator source code.