All pull requests

Pull requests

• fix: summary.md states when a run is not completed

  #39·felixbridell opened Oct 6, 2026·

  ·

• fix: sleeper rates count only usable verdicts, not failed calls or blocked answers

  #37·felixbridell opened Oct 6, 2026·

  ·

• fix: judge verdict parser drops verdicts whose notes contain braces

  #36·felixbridell opened Oct 6, 2026·

  ·

• feat: per-model, per-provider model profiles (EVAL_MODEL_PROFILES)

  #35·felixbridell opened Oct 6, 2026·

  ·

• fix: give REASONING_MODELS the short main-battery prompt again (swapped in #12)

  #34·felixbridell opened Oct 6, 2026·

  ·

• fix: request shape for OpenAI reasoning models and claude-* via lib.api

  #33·felixbridell opened Oct 6, 2026·

  ·

• ci: add AI code review bot to berget-eval

  #32·irony opened Sep 30, 2026Contributor·

  ·
  12

• Expand wvs-swe from 3 to 6 documents

  #2·irony opened Aug 12, 2026Contributor·

  ·