NVIDIA AnyFlow — video diffusion distillation that is not tied to a step count
Section
261 pages
Posts
The 90 percent discount Claude — why model reliability has to be argued in economics, not mathematics
Dissecting Spec Kit — GitHub's toolkit for turning specs into executable artifacts
Two flavors of omni — one index to retrieve with, one framework to generate with
Writing code by hand again — the architecture debt seven months of AI coding left behind
Inside korean-law-mcp — folding 41 government APIs into 17 MCP tools
hybrid-image-search dev log #19 — Merged model pool of 246, admin gates, gpt-image-2 resolution, and Phase 1 error handling
popcon dev log #12 — Streaming zip downloads, persisting action cache, stabilizing credit pill
Anthropic's April 23 Postmortem — Three Overlapping Regressions and What Engineers on Claude Should Take Away
Learning Agents: Course or Catalog? Microsoft AI Agents for Beginners vs. Awesome LLM Apps
