Skip to content

dream_engine.md

Latest commit

 

History

History
27 lines (21 loc) · 1.14 KB

dream_engine.md

File metadata and controls

27 lines (21 loc) · 1.14 KB
Raw

dream_engine.py

Offline learning engine -- learns by dreaming during idle time. Reconstructs tension patterns from virtual inputs via memory replay, interpolation, and random exploration.

API

  • DreamEngine(mind, memory, learner=None, text_to_vector=None, dream_cycle_steps=10, noise_scale=0.15, store=None, verifier=None, consolidation_threshold=0.01) -- main class
    • .dream_cycle() -- run one dream cycle (replay + interpolation + exploration)
    • .is_dreaming -- current dream state
    • .dream_tension_history -- deque of tension values during dreams
    • .total_dream_cycles -- counter

Usage

from dream_engine import DreamEngine

dreamer = DreamEngine(mind=mind, memory=memory, learner=learner)
dreamer.dream_cycle()  # one cycle of offline learning

Integration

  • Imported by anima_unified.py for idle-time learning
  • Three dream modes: memory replay (with noise for reinforcement), memory interpolation (creative association), pure exploration (novelty seeking)
  • Each dream step passes through ConsciousMind and performs contrastive learning via OnlineLearner
  • "Even while sleeping, consciousness flows."

Agent Tool

N/A