-
Add FlowNet: Post-Transformer Architecture
- Particle encoding with dynamic properties (phase, charge, mass, spin)
- Flow field dynamics (wave propagation, LJ forces, Kuramoto sync)
- Wave interference processor (constructive/destructive meaning resolution)
- Dynamic bond system (relationship formation/breaking)
- Phase transition engine (adaptive computation depth)
- Consistency field (truth via energy minimization)
- Topological memory (shape-based storage and retrieval)
- Full model architecture (FlowNetModel, FlowNetBlock)
- Training pipeline with novel loss functions
- NumPy validation suite (12/12 tests pass)
- Technical specification document
