Agent evaluation workshop: a perfect single-run score hides unreliability
AI Engineer · youtube · 2026-10-12
At AI Engineer, Weights & Biases' Wolfram Ravenwolf presented a full agent evaluation workshop. An agent scoring perfectly on one git-recovery task stays off leaderboards — WolfBench tracks average/best/worst runs and tasks solved once vs. every time across six pipeline layers (benchmark, runner, sandbox, agent, model service, settings). Key traps: cheaper models can cost more via token use, and higher reasoning effort can lower scores by eating execution time. Live stack: Harbor runs Terminal Bench, Marimo filters runs, Weave preserves trajectories.
More from coding & agent
- OpenJet claims Qwen3.8-27B coding runs 1.9x faster than llama.cpp on one RTX 3090 — JeremyCMorgan · 2026-10-23
- Building an autonomous SEO agent in Hermes with three skills: programmatic pages, Keywords Everywhere, DataForSEO — EXM7777 · 2026-10-23
- Building a sports betting AI agent with onchain execution and human approval by default: architecture and lessons — Dry-Comfortable-2678 · 2026-10-23
- Capy 0.5.0 lets Claude Pro/Max subscribers reuse their plan quota, adds OIDC cloud access — garrytan · 2026-10-23
- You.com CEO Richard Socher: legacy code is human slop, let AI clean it up — RichardSocher · 2026-10-23
- Let agents design a single-line glance component so you can track them from the sidebar — blixt · 2026-10-23