He Ran a CNN Inside a PDF: 26,698 Parameters, No Python, Pure Acrobat JS
cneuralnetwork · x · 2026-11-07
A developer packed a PyTorch-trained MNIST digit classifier directly into a 546 KB PDF: 26,698 learned parameters, two conv layers, and JavaScript that runs the forward pass. Once opened in Acrobat Reader, prediction needs no Python, browser, API, or network connection — and it runs quite accurately.
Related event: Developer Runs a CNN Inside a PDF File(2 posts)→
More from coding & agent
- Ahrefs Doubles API Unit Limits Across All Tiers for Agent Builders — gaganghotra_ · 2026-11-07
- The AI agent to-do list meme: fix a bug, solve the Riemann hypothesis, cure cancer — fofrAI · 2026-11-07
- Browser MCP vs Playwright MCP: Reddit debate on when agents need browser tools at all — TwoBlueFools · 2026-11-07
- A 4B router model beats fixed Opus 5.5 on Terminal-Bench, 77.78 vs 69.44 — YangYou1991 · 2026-11-07
- Testing and verification infra is quietly becoming a strategic advantage in the AI era — arpit_bhayani · 2026-11-07
- Indie dev shares AI-generated action RPG concept with top-down and first-person modes — Dimillian · 2026-11-07