Module gradient_checkpointing 

Gradient Checkpointing

Implements gradient checkpointing to reduce memory usage during training:

• Selective activation storage
• Recomputation during backward pass
• Memory-efficient training for large models
• Configurable checkpoint intervals

Structs

CheckpointConfig

Gradient checkpointing configuration

CheckpointManager

Checkpoint manager

CheckpointStats

Checkpoint statistics

CheckpointedLayer

Checkpointed layer wrapper

CheckpointedSequential

Sequential model with gradient checkpointing

Enums

CheckpointStrategy

Checkpoint strategy

Functions

estimate_memory_savings

Utility function to estimate memory savings