Articles tagged: "graph-neural-networks"

• Aug 24, 2026

  The Plateau: Five Ways to Enrich a Heatmap, Zero Breakthroughs

  The last post in this project ended with a promise. Eradicating Syntax walked through a Diffusion Transformer that generates executable graphs instead of code, and closed with “What’s next: RLAIF.”...
• May 5, 2026

  Eradicating Syntax: Building a Neural Universal Machine That Executes Graphs, Not Code

  Our GNN autoencoders achieved 81% node accuracy on Ruby ASTs yet produced 0% valid code. The culprit was the literal value bottleneck — nearly half of every AST consisted of...
• Apr 20, 2026

  The Literal Value Bottleneck: Why GNN Autoencoders Fail at Code Generation

  We trained GNN autoencoders on 22,000 Ruby ASTs. The models achieved 81% node type accuracy and 99.5% type diversity, yet generated exactly 0% syntactically valid code. Here is why.