Arxis Package
adaptive-memory
Learned memory management for ML inference: predictive prefetching, adaptive compression, and intelligent caching.
adaptive-memory replaces fixed cache policies with a neural system that learns access patterns and optimises cache placement and compression per item.
- LSTM-based access pattern prediction
- Importance-aware compression selection
- Automatic promotion across L1/L2/L3 cache tiers
- Prefetching before access demand
pip install adaptive-memory