HASH researchers detail the technical challenges of inverse problems in agent-based models, which aim to recover hidden mechanisms from observed data, and argue that solving these problems is essentia…
HASH engineers have determined that embeddings in knowledge graphs should be treated as encodings of entities rather than metadata, requiring content-grade access control and metadata-grade lifecycle …
HashQL, a hypergraph query language supporting multiple independent time dimensions, was released as a research preview. Developed by the hgres team, it combines temporality, typing, and graph structu…
HASH researchers argue that building a process foundation model (PFM) requires a novel architecture, Dynamic Graph-JEPA, which extends Graph-JEPA to handle partially-observed, typed, temporal graph st…
Researchers at HASH built a 25M-parameter transformer decoder, vec2slug, that generates URL slugs directly from vector embeddings, achieving 14-19x faster and 85x cheaper performance than a Haiku-clas…