-
Notifications
You must be signed in to change notification settings - Fork 0
Ingestion
devlux76 edited this page Mar 14, 2026
·
1 revision
This page describes how CORTEX ingests new observations and integrates them into memory.
- Chunking/Parsing — Raw inputs are segmented into pages/blocks.
- Embedding — Each chunk is embedded using a Matryoshka-capable model.
- Fast Neighbor Insert — New vectors are connected into the semantic neighbor graph.
- Hierarchical Prototypes — Pages are organized into Books, Volumes, and Shelves.
CORTEX manages a hierarchical prototype structure to keep hot (frequently-accessed) concepts in memory while relying on disk-backed storage for the long tail.
For implementation details, see the code in
hippocampus/and theHierarchyBuilderdesign notes.