Issues

• Request leaderboard entry: KDL-Frontier-Parser-nano-v2

  run-llama/ParseBench#225 · Filter by author skihyeon opened on Oct 28, 2026

• parse-bench run exits 0 when a category evaluation fails, and Ctrl-C only skips to the next category

  run-llama/ParseBench#219 · Filter by author RaphaelFakhri opened on Oct 25, 2026

• Chart rules for ADL p17 (Figure 10) use axis-side annotations as series labels, failing correct single-series extractions

  run-llama/ParseBench#215 · Filter by author zmtomorrow opened on Oct 23, 2026

• Request for evaluation assistance

  run-llama/ParseBench#209 · Filter by author AlamoTNT opened on Oct 14, 2026

• Verify evals on Papers with Code

  run-llama/ParseBench#207 · Filter by author NielsRogge opened on Oct 13, 2026

• Run and publish the full Apple Vision benchmark

  run-llama/ParseBench#175 · Filter by author miqcie opened on Sep 22, 2026

• Unfiltered evaluation drops text_content cases sharing IDs with text_formatting

  run-llama/ParseBench#173 · Filter by author miqcie opened on Sep 22, 2026

• Open-weight leaderboard rows are not reproducible: the markdown post-processing used by the self-hosted providers is unspecified

  run-llama/ParseBench#170 · Filter by author AlamoTNT opened on Sep 22, 2026

• OrcaRouter provider support for ParseBench

  run-llama/ParseBench#145 · Filter by author putraperdana1207-pixel opened on Sep 10, 2026

• RuleBasedMetric's signal.alarm timeout fails off the main thread (ParseEvaluator from a thread pool)

  run-llama/ParseBench#144 · Filter by author jpl42 opened on Sep 10, 2026
  1

• Proposal: an EvalPort adapter so ParseBench results are portable outside this repo's own report format

  run-llama/ParseBench#110 · Filter by author adhabnr-ux opened on Aug 25, 2026

• is_bold accepts any heading line containing the text, so heading promotion earns bold credit — a ~20-line transformation exceeded the then-leading score (disclosed, not exploited)

  run-llama/ParseBench#104 · Filter by author ammoman21 opened on Aug 19, 2026

• ChartDataPointRule rejects correct data extraction when the chart title is not bold or a heading

  run-llama/ParseBench#100 · Filter by author ammoman21 opened on Aug 19, 2026

• Share raw per-document outputs alongside scores

  run-llama/ParseBench#38 · Filter by author hashiromer opened on May 22, 2026
  3