This directory consolidates all documentation about model architectures and training systems used in Trinity.
- JEPAT/ - Ternary Joint Embedding Predictive Architecture
- NCA/ - Neural Cellular Automata
- VSA/ - Vector Symbolic Architecture
- Ternary/ - Ternary computing and representation
- Hybrid/ - Hybrid BigInt and arithmetic
- Architecture - TrinityBlock, masks, EMA, MSE loss
- Parameters - Training configuration and multipliers
- Experiments - Experimental results (J-000, J-001)
- Architecture - Grid configuration, states, entropy
- Entropy Bands - Wave 8.5 G1-G8 sweep
- Integration - Multi-objective with JEPA/NTP
- Overview - Core VSA concepts and FPGA implementation
- Operations - Quick reference for bind/unbind/bundle
- API Reference - Links to complete API docs
- Balanced Ternary - Complete ternary guide (link)
- Representation ADR - Packed trit encoding (link)
- API Reference - HybridBigInt API (link)
- v2.0 Report - Implementation report from gh-pages
- v2.1 Report - Latest improvements report
- HSLM Training: ../../experiments/FOUND_EXPERIMENTS_SUMMARY.md
- SEVO Method: ../../lab/papers/sevo-method.md
- Framework: ../../research/TRINITY_S3AI_UNIFIED_FRAMEWORK.md
- Glossary: ../../glossary.md
Last updated: 2026-04-24