Module reasoning

reasonkit::evaluation

Module reasoning 

Source
Expand description

§Reasoning Quality Metrics

Core metrics for evaluating AI THINKING improvement with ReasonKit.

This is the HEART of ReasonKit evaluation - measuring whether ThinkTool protocols actually improve AI reasoning.

Structs§

BedRockMetrics
BedRock-specific metrics
BenchmarkResult
Result from running a benchmark
BrutalHonestyMetrics
BrutalHonesty-specific metrics
CalibrationMetrics
Calibration metrics (confidence vs accuracy)
ConsistencyMetrics
Self-consistency metrics
GigaThinkMetrics
GigaThink-specific metrics
LaserLogicMetrics
LaserLogic-specific metrics
ProofGuardMetrics
ProofGuard-specific metrics
QuestionResult
Result for a single question
ReasoningMetrics
Aggregated reasoning metrics
ThinkToolMetrics
Generic ThinkTool effectiveness metrics

Enums§

Profile
ReasonKit profiles (ThinkTool chains)
ThinkTool
Individual ThinkTools

Functions§

calculate_thinktool_delta
Calculate improvement delta for a ThinkTool
is_significant
Statistical significance test (simplified)