Module quantization 

Advanced quantization support for model compression and acceleration.

This module provides comprehensive quantization capabilities including:

• Multiple quantization schemes (INT8, INT4, FP8, binary)
• Quantization-Aware Training (QAT) support
• Post-Training Quantization (PTQ) with calibration
• Per-channel and per-tensor quantization
• Symmetric and asymmetric quantization
• Dynamic and static quantization modes
• Quantization simulation for accuracy validation

Structs

CalibrationStats

Statistics collected during calibration.

FakeQuantize

Fake quantization for QAT (simulates quantization during training).

NodeId

Node identifier (0-based index into graph.nodes).

QuantizationConfig

Quantization configuration for a graph or model.

QuantizationParams

Quantization parameters for a tensor.

QuantizationSummary

Summary of quantization results.

Quantizer

Quantizer for converting graphs to quantized representations.

Enums

CalibrationStrategy

Calibration strategy for post-training quantization.

QuantizationError

Quantization-related errors.

QuantizationGranularity

Quantization granularity.

QuantizationMode

Quantization mode.

QuantizationSymmetry

Quantization symmetry mode.

QuantizationType

Quantization data types.