FROM THE AUTHOR
Explore the published guides and contributions from John C. Mullen.
88 published articlesA review rubric converts subjective approval into two machine-readable outputs: a score per quality axis and a named rejection reason when a draft fails. This guide explains…
GGUF, Safetensors and GPTQ are not three versions of the same thing. GGUF and Safetensors are container formats, while GPTQ is a quantisation method whose weights are…
Vendor and analyst guidance converges on the same default: start with one agent and add more only when a specific failure demands it. This guide explains what…
A clarify-before-answer gate is an explicit prompt rule that tells a model to ask, declare its assumptions, or stop when a request is missing something it needs.…
A WordPress AI content style guide turns brand voice into explicit rules that AI tools can follow. Start with the 3 Cs—Character, Consistency, Context—then encode tone, image…
Start with Q4_K_M, then let the model repository override that default. This guide explains what GGUF quantization actually changes, how to read the documented quality, size, and…
Prompt caching and prefix caching both reuse the key-value (KV) state a model already computed for an unchanged prompt prefix, so a new request can skip that…
Secrets leak into RAG when ingestion pipelines index everything without classification, scanning, or access metadata. The fix is layered: keep credentials out of source documents, scan and…
Function calling lets a local model request a specific function by name and arguments, while your application decides whether and how to run it. This guide explains…
Setting stream: true on an AI API request is only half the work. This guide explains how to enable streaming, parse server-sent events without corrupting split chunks,…
A trace ID strategy is a small contract: one trace ID per unit of autonomous work, one span ID per step, a parent link for every child,…
A RAG pipeline can fail in two different places, and one end-to-end score will not tell you which. Test retrieval and generation separately: score Precision@k and Recall@k…