Engineer Finds 11 Confident Errors All Came From His Toolchain, Not the LLM

Engineer Colin McNamara recounts how his local inference stack delivered 11 confident errors, none caused by the LLM itself—recent ones traced to a cache-busting bug that made a 24x performance gap appear as parity.

2026-10-13 ~ 2026-10-13 · 2 related posts

• Cache-busting bug masked a 24x GPU gap: 11 confident falsehoods, none from the model — colinmcnamara · 2026-10-13
• "It Was Never the Model": 11 Falsehoods in a Local Inference Stack, All From Tooling — colinmcnamara · 2026-10-13