AI article
Only the Design of a Local LLM Eval Harness Survives — Rebuilding It for Qwen3.8 27B
Community description: Three weeks ago I wrote an RFC 6902 porting harness to decide whether to keep Ornith. I reused it to evaluate Qwen3.8 27B — except the script itself had vanished with /tmp, and rebuilding took minutes only because the design was written down. Plus a record of the two predictions it proved wrong.
Dev.to | Nov 9, 2026 | kohei_wada
Automated excerpt
It isn't a benchmark — it's actual porting work. That's split into four phases, run one phase per session, with the model unloaded between phases so nothing carries over. So "I reused the same script" isn't accurate.
Selected automatically from source text; not independently written or fact-checked. Read the original for full context.