All pull requests

Pull requests

• fix(validator): return submission evaluations in chronological order

  #256·tryeverything24 opened Jul 24, 2026·

  ·

• fix(fugu): accept a bare integer access index in workflow parse-gate

  #255·rsnetworkinginc opened Jul 24, 2026·

  ·
  1

• fix(reward): extract \boxed{} before grading RLPR WebInstruct math (#170)

  #254·minion1227 opened Jul 24, 2026Contributor·

  ·
  1

• Require the checker to complete in run_pass_at_1 (block early-exit reward hack)

  #253·boskodev790 opened Jul 24, 2026·

  ·

• fix(validator): commit evaluation progress so it is visible while the run is in flight (#251)

  #252·kai392 opened Jul 24, 2026·

  ·
  1

  2

• fix(validator): failed re-evaluation inherits the previous run's score from a stale results.json (#249)

  #250·kai392 opened Jul 24, 2026·

  ·
  1

  2

• fix(worker): one-shot run exits 0 after a successful pass

  #248·marktech0813 opened Jul 24, 2026·

  ·
  1

• fix(eval): cancel queued provider-eval jobs when a standalone evaluation is deleted

  #247·tryeverything24 opened Jul 24, 2026·

  ·

• Skip Verifier turns when selecting the committed answer

  #246·boskodev790 opened Jul 23, 2026·

  ·

• Fall back to reasoning_content when model content is empty

  #245·boskodev790 opened Jul 23, 2026·

  ·

• fix(cost): serialize ledger records with json.dumps so ids can't corrupt JSON (#165)

  #241·minion1227 opened Jul 23, 2026Contributor·

  ·
  1

  1

• fix(fugu): accept a bare-int access index in the workflow parser (#225)

  #240·minion1227 opened Jul 23, 2026Contributor·

  ·
  1

  1

• fix(reward): widen multiple-choice alphabet to A-J so MMLU-Pro E-J grades (#116)

  #239·minion1227 opened Jul 23, 2026Contributor·

  ·
  2

  2

• fix(validator): guard legacy ensure_schema alters for fresh DBs (#120)

  docsevalvalidator
  #230·RealDiligent opened Jul 20, 2026Contributor·

  ·
  1

  3

• fix(scripts): default remote GPU helpers to index 5 (#33)

  docsevaltrain
  #229·RealDiligent opened Jul 20, 2026Contributor·

  ·
  1

  3

• fix(fugu): accept a bare-int access index in workflow parsing

  docsminertrain
  #226·joaovictor712 opened Jul 17, 2026Contributor·

  ·
  1

• fix(dataset): fail loudly instead of silently scoring on the toy set (#65)

  benchmarkdocsminertrain
  #224·jeffrey701 opened Jul 17, 2026·

  ·
  1

• fix(cost): serialize ledger rows as JSON

  docsminertrain
  #221·Steve-too opened Jul 16, 2026·

  ·
  1

• Fix #189: apply per-role decoding config in run_trajectory

  train
  #220·marktech0813 opened Jul 16, 2026·

  ·
  1

• fix(reward): bipartite-match BFCL multi-call grading (#180)

  docsminertrain
  #218·kiannidev opened Jul 16, 2026Contributor·

  ·
  1

• fix(reward): grade committed MC letter, not first echoed option (#124)

  docsminertrain
  #217·kiannidev opened Jul 16, 2026Contributor·

  ·
  1

• fix(reward): strip LaTeX-escaped set braces without trailing backslash (#162)

  docsminertrain
  #216·kiannidev opened Jul 16, 2026Contributor·

  ·
  1

• feat(diagnostics): per-question agreement + contested-subset export (#60)

  docsevalminertrain
  #215·jeffrey701 opened Jul 15, 2026·

  ·
  1

• fix: correct off-by-one in load_project_env default repo_root

  minertrain
  #213·lourincedaging0-commits opened Jul 15, 2026·

  ·

• fix(benchmarks): make rlpr/ifeval/bfcl_simple load() callable instead of raising TypeError (#152)

  benchmarkdocsminer
  #179·joaovictor91123 opened Jul 13, 2026Contributor·

  ·
  1