Module gradient_debugger 

Gradient Debugging and Analysis Tools

This module provides comprehensive gradient debugging capabilities including real-time monitoring, anomaly detection, performance tracking, conflict analysis, visualization, and enhanced analysis for gradient flow in neural networks.

Architecture

The gradient debugging system is organized into several focused modules:

• types - Core types, enums, and configuration for gradient debugging
• monitoring - Real-time gradient monitoring and adaptive thresholds
• anomaly_detection - Advanced gradient anomaly detection system
• performance_tracking - Performance tracking and bottleneck analysis
• conflict_analysis - Gradient conflict analysis between layers
• visualization - Gradient flow visualization data generation
• enhanced_analysis - Enhanced layer analysis and network-level insights
• debugger - Main GradientDebugger orchestrating all components

Re-exports

pub use debugger::GradientDebugger;

pub use monitoring::AdaptiveThresholds;

pub use monitoring::MonitoringConfig;

pub use monitoring::MonitoringResults;

pub use monitoring::RealTimeGradientMonitor;

pub use anomaly_detection::AnomalyContext;

pub use anomaly_detection::AnomalySummary;

pub use anomaly_detection::AnomalyTrend;

pub use anomaly_detection::AnomalyType;

pub use anomaly_detection::BaselineGradientStats;

pub use anomaly_detection::GradientAnomaly;

pub use anomaly_detection::GradientAnomalyDetector;

pub use performance_tracking::GradientPerformanceTracker;

pub use performance_tracking::LayerPerformanceMetrics;

pub use performance_tracking::OptimizationIssue;

pub use performance_tracking::OptimizationRecommendation;

pub use performance_tracking::OptimizationSeverity;

pub use performance_tracking::PerformanceSnapshot;

pub use performance_tracking::PerformanceTimer;

pub use performance_tracking::PerformanceTrends;

pub use performance_tracking::ResourceUtilization;

pub use conflict_analysis::ConflictLevel;

pub use conflict_analysis::ConflictMitigationStrategy;

pub use conflict_analysis::ConflictReport;

pub use conflict_analysis::ConflictType;

pub use conflict_analysis::GradientConflict;

pub use conflict_analysis::GradientConflictAnalysis;

pub use conflict_analysis::MitigationComplexity;

pub use visualization::CriticalGradientPath;

pub use visualization::ExplodingRegion;

pub use visualization::FlowEdge;

pub use visualization::FlowNode;

pub use visualization::GradientDeadZone;

pub use visualization::GradientFlowNetwork;

pub use visualization::GradientFlowVisualization;

pub use visualization::GradientLayerFlow;

pub use visualization::GradientVisualizationConfig;

pub use visualization::TemporalGradientFlow;

pub use visualization::VanishingRegion;

pub use enhanced_analysis::EnhancedLayerGradientAnalysis;

pub use enhanced_analysis::GradientHierarchy;

pub use enhanced_analysis::LayerGradientDetails;

pub use enhanced_analysis::LayerOptimizationSuggestion;

pub use enhanced_analysis::NetworkLevelAnalysis;

pub use enhanced_analysis::OptimizationPriority;

pub use debugger::ComprehensiveGradientReport;

pub use debugger::GradientDebugStatus;

pub use debugger::GradientRecommendation;

pub use debugger::LayerGradientStatus;

pub use debugger::PerformanceInsights;

pub use debugger::RecommendationType;

pub use types::*;

Modules

anomaly_detection

Advanced Gradient Anomaly Detection System

conflict_analysis

Gradient Conflict Analysis Between Layers

debugger

Main Gradient Debugger Implementation

enhanced_analysis

Enhanced Layer Analysis and Network-Level Insights

monitoring

Real-time Gradient Monitoring and Adaptive Thresholds

performance_tracking

Performance Tracking and Bottleneck Analysis for Gradient Computation

types

Core Types and Configuration for Gradient Debugging

visualization

Gradient Flow Visualization Data Generation

Enums

RecommendationPriority

Priority levels for recommendations

Type Aliases

GradientDebugReport