Expand description
Memory-efficient optimizer implementations
This module provides optimizers designed to minimize memory usage during training, particularly useful for large models or resource-constrained environments.
Structsยง
- Circular
Buffer - Circular buffer for L-BFGS history with memory management
- Memory
Config - Configuration for memory-efficient optimizers
- Memory
Efficient Adam - Memory-efficient Adam optimizer with reduced memory footprint
- Memory
EfficientLBFGS - Memory-efficient L-BFGS with improved history management
- Memory
Efficient Optimizer Builder - Builder for memory-efficient optimizers
- Memory
Pool - Memory pool for reusing tensor allocations