AI article
Temperature 0, same model, two different GPUs: find the word that splits them, then pin it
Community description: Greedy decoding on the same GGUF still flips an answer between two cards. Here is a script that finds the exact token where two runs part, the one stock setting that stops it, and a CI gate that keeps a reproducibility claim honest.
Dev.to | Nov 8, 2026 | Dekita
Automated excerpt
So if two runs of the same build on the same card agree and two runs on different cards do not, the variation is upstream of the sampler. The engine computed two slightly different probability tables, and greedy decoding faithfully picked from two slightly different tables. Same card, same build, same settings hash, then the numbers are comparable.
Selected automatically from source text; not independently written or fact-checked. Read the original for full context.