Skip to content

claw-mem v3.3.0

Choose a tag to compare

View all tags
@petercheng petercheng released this 24 May 04:43
· 323 commits to main since this release
v3.3.0
e7e2506

claw-mem v3.3.0 — Dual System Memory

Summary

Hippocampus-Neocortex inspired dual memory system for rapid learning and long-term consolidation.

What's New

🧠 Dual System Memory Model

HippocampalStore (Fast) → ConsolidationLoop → NeocorticalStore (Slow)

HippocampalStore — Fast, Short-Term Memory

  • Rapid encoding (< 1ms)
  • Short-term memory storage (default 24h TTL)
  • LRU cache for hot access optimization
  • Importance tagging for consolidation priority

NeocorticalStore — Slow, Long-Term Memory

  • Concept abstraction from memories
  • Long-term consolidation
  • Ebbinghaus forgetting curve
  • Semantic connections between concepts

ConsolidationLoop — Periodic Memory Consolidation

  • Background thread execution (non-blocking)
  • Importance-weighted consolidation
  • Forgetting curve application
  • Automatic hippocampal cleanup after consolidation

Performance

Operation Latency
Hippocampal Store < 1ms
Hippocampal Retrieve < 10ms
Neocortical Consolidate ~50ms

Quick Start

from claw_mem.dual_system import (
    HippocampalStore, NeocorticalStore, ConsolidationLoop,
    Memory, Concept, DualSystemConfig,
)

hippo = HippocampalStore()
cortex = NeocorticalStore()
loop = ConsolidationLoop(hippo, cortex)

# Fast storage
memory_id = hippo.store(Memory(content="Important insight", importance=0.8))

# Start background consolidation
loop.start_background()

Tests

  • 37 tests passed
  • Comprehensive unit tests for all dual system components

Full Changelog

See CHANGELOG.md

Assets 2
Loading