Expand description
Computation graph analysis tools for debugging deep learning models.
This module provides comprehensive analysis tools for computation graphs, including node analysis, dependency tracking, optimization opportunities, bottleneck detection, and graph visualization capabilities.
Structs§
- Bottleneck
Analysis - Bottleneck analysis results
- Complexity
Analysis - Complexity analysis of the computation
- Computation
Graph - Represents a computation graph for analysis
- Computation
Graph Analyzer - Main computation graph analyzer
- Data
Flow Analysis - Data flow analysis results
- Estimated
Improvement - Estimated improvement from an optimization
- Flop
Analysis - FLOP (Floating Point Operations) analysis
- Graph
Analysis Config - Configuration for computation graph analysis
- Graph
Analysis Result - Comprehensive analysis result for a computation graph
- Graph
Metadata - Metadata about the computation graph
- Graph
Node - Represents a single node in the computation graph
- Graph
Statistics - Graph statistics
- Memory
Analysis - Memory usage analysis
- Memory
Reuse Opportunity - Memory reuse opportunity
- Optimization
Opportunity - Optimization opportunity detection
- Variable
Lifetime - Lifetime information for a variable
Enums§
- Operation
Type - Types of operations in the computation graph
- Optimization
Priority - Priority levels for optimizations
- Optimization
Type - Types of optimizations that can be applied