All pull requests

Pull requests

• fix(locomo): correct category mapping to match official dataset

  #90·ramblinghermit0403 opened Aug 27, 2026·

  ·

• fix(ingest): persist document ids so a mid-question failure doesn't lose them

  #88·Agnik47 opened Aug 20, 2026·

  ·
  1

• fix(benchmarks): make question order deterministic and load() idempotent

  #87·Agnik47 opened Aug 14, 2026·

  ·
  1

• fix(sampling): use an unbiased, seedable shuffle for random question selection

  #86·Agnik47 opened Aug 14, 2026·

  ·
  1

• fix(checkpoint): report save failures instead of dropping or crashing on them

  #85·Agnik47 opened Aug 14, 2026·

  ·
  1

• fix(models): reject unknown model aliases instead of assuming OpenAI

  #84·Agnik47 opened Aug 14, 2026·

  ·
  1

• fix(rag): guarantee forward progress in chunkText

  #83·Agnik47 opened Aug 14, 2026·

  ·
  1

• fix(longmemeval): judge abstention questions with the abstention rubric

  #82·thegoodengineer opened Aug 14, 2026·

  ·

• fix(rag): make BM25 indexing incremental and idempotent

  #81·Agnik47 opened Aug 14, 2026·

  ·
  1

• Fix BEAM benchmark correctness and reproducibility

  #58·ved015 opened Aug 6, 2026·

  ·
  2

• fix(providers): rate-limit resilience + non-fatal ingest failures

  #57·dkoosis opened Aug 2, 2026·

  ·

• Add LongMemEval-V2 benchmarking to MemoryBench

  #55·ved015 opened Jul 28, 2026·

  ·
  1

• Add InfiniteBench benchmark adapter with dataset loading workflow

  #54·aijadugar opened Jul 28, 2026·

  ·
  

• Add controlled LongMemEval-V2 small ingestion workflow

  #50·ved015 opened Jul 17, 2026·

  ·

• Add BABILong benchmark adapter

  #49·aijadugar opened Jul 16, 2026·

  ·
  2

  

• feat(providers): add Basic Memory provider

  #47·groksrc opened Jun 11, 2026·

  ·
  1