Module optimization_history 

TensorOptimizationHistory — records and analyzes optimization steps (loss/gradient values) to detect convergence, track best results, and guide adaptive learning rate schedules.

Structs

HistoryStats

Aggregated statistics computed over the entire recorded history.

OptimizationHistoryConfig

Configuration for TensorOptimizationHistory.

OptimizationStep

A single recorded step in the optimization process.

TensorOptimizationHistory

Records and analyzes a history of optimization steps.

Enums

ConvergenceStatus

The convergence state inferred from recent optimization history.