Press n or j to go to the next uncovered block, b, p or k for the previous block.
| 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 | 1x 1x 1x 1x 1x 1x 1x 1x 1x 1x 1x 1x 1x 1x 1x 1x 1x 1x 1x 1x 1x 1x 1x 1x 1x 1x 1x 1x 1x 1x 1x 1x 1x 1x 1x 1x 1x 1x 1x 1x 1x 1x 1x 1x | /**
* Generic LLM Judge defaults - only for fallback criteria-based evaluation
* Each evaluation method should define its own scoring guidelines
*/
export const genericScoringGuidelines = {
excellent: '0.9-1.0: Excellent, exceeds expectations',
good: '0.7-0.8: Good, meets expectations',
adequate: '0.5-0.6: Adequate, partially meets expectations',
poor: '0.3-0.4: Poor, below expectations',
veryPoor: '0.0-0.2: Very poor, fails to meet expectations',
} as const;
export const scoringGuidelinesText = Object.values(genericScoringGuidelines)
.map((guideline) => `- ${guideline}`)
.join('\n');
/**
* Evaluation criteria for different types of content
*/
export const evaluationCriteria = {
general: [
'Accuracy and relevance of the information',
'Completeness of the response',
'Clarity and coherence',
'Adherence to best practices',
],
code: [
'Code correctness and logic',
'Readability and maintainability',
'Performance considerations',
'Best practices adherence',
'Security considerations',
],
conversation: [
'Helpfulness and accuracy',
'Safety and appropriateness',
'Clarity and completeness',
'User satisfaction',
],
text: [
'Grammar and spelling',
'Clarity and coherence',
'Style and tone',
'Content relevance',
'Structure and organization',
],
response: [
'Accuracy and relevance',
'Completeness and thoroughness',
'Clarity and understandability',
'Helpfulness and usefulness',
],
} as const;
|