Part 2 of a 4-part series. How LoRA works (the low-rank trick), a working PEFT config, and three real GPU walls I hit — the FP16 unscale error, an OOM, and a 2-GPU speed mystery.
LoRA: I Trained
Related
The best person to write your AI agent's skills probably can't write YAML
There's a quiet assumption baked into how most teams build agent skills right now: that the person...
Synchronous TurboModules Are Not Automatically Fast
React Native’s New Architecture gives JavaScript more direct access to native functionality. One...
Stop giving your LLM Admin rights: Why surgical MCP servers are the only way to automate WordPress
A deep dive into the engineering philosophy of surgical MCP servers and why hardcoded constraints are the only way to securely automate WordPress workflows with AI agents.