staticwebpage

Meet ADM, Genefold AI's proprietary non-autoregressive decision runtime.

A protocol for combining ArrowSpace local basin analysis with epiplexity to distinguish trivial smooth regions from useful learnable structure inside transformer subspaces.

A corrected derivation of diffusion under a metric that blends raw item geometry with an ArrowSpace feature-manifold projection

Comparative semantic probing: dual-space analysis on transformer weights and embeddings

Notebook walkthrough: using Rayleigh energy as a landscape signal for high-dimensional embedding spaces

Compare 384-dim embeddings with the new 1024-dim embeddings

Testing all aspects of Graph Wiring on semantic data — why geometry fails, measuring the win with MRR-Top0

Testing all aspects of Graph Wiring on semantic data — how fast and accurate arrowspace is

Training Graph Wiring on the Dorothea dataset — speed, applications, and the connection between Rayleigh score and epiplexity

An hypothesis for arrowspace as minimal surface approximator

Training notes for a 30M-class TauGPT run including the data pipeline, validation routing, and stabilization knobs

Domain memory injected directly inside self-attention via a persistent Graph Laplacian

AI safety through topology-aware, energy-informed retrieval that separates stable facts from risky intuitions

Test-bed milestone for a unified vector, graph, and key-value engine built on spectral indexing

Milestone release completes the search-matching-ranking pipeline with stabilized energymaps module

Deep dive into a Rust implementation of a decoder-only transformer inspired by Karpathy's nanoGPT

Define a new score that generalizes MRR to the entire top-k by combining list metrics with topology signals

Rust implementation of DeepSeek-OCR compression achieves 10x token reduction

Version 0.16.0 is out with encouraging results for arrowspace as one of the fastest ANN algorithms available in the open

Evaluation on a CVE corpus spanning 1999 to 2025 shows spectral modes preserve head agreement while enhancing long-tail relevance

This release rethinks how arrowspace builds and queries graph structure from high-dimensional embeddings