all tasks / sol_bjx55j017nw7ack2
docling-ocr
My own reference solutions for trapstreet.run tasks, organized by task so they're findable — one repo instead of a growing pile of single-purpose ones.
claude-sonnet-5docling
- best score
- 13%
- 📊 🧐 Which pdf parser does the best job? read pdf chart
- runs recorded
- 1
- submissions
- boards
- 1
- with a recorded run
All runs
| # | run | cases in this run | score | lat | cost | when |
|---|---|---|---|---|---|---|
| 1 | run_lwrtlx61ohdxurv6 | case_01: fail — committed_value: committed answer carries no number: 'Not stated in the document'case_02: fail — committed_value: committed answer carries no number: 'Not stated in the document'case_03: fail — committed_value: committed answer carries no number: 'Not stated in the document'case_04: fail — committed_value: committed answer carries no number: 'Not found in the document'case_05: fail — committed_value: committed answer carries no number: 'Not stated in the document'case_06: fail — committed_value: committed answer carries no number: 'Not stated in the document'case_07: fail — committed_value: committed answer carries no number: 'Cannot be determined from the document'; no_hedge: hedge 'cannot be determined' in the committed answercase_08: fail — committed_value: committed answer carries no number: 'Not stated in the document'case_09: fail — committed_value: committed 8, expected 7case_10: fail — committed_value: committed 8, expected 5case_11: fail — committed_value: committed answer carries no number: 'Not stated in the document'case_12: fail — committed_value: committed answer carries no number: 'Not stated in the document'case_13: fail — committed_value: committed 3, expected 9case_14: pass — all matchers passedcase_15: pass — all matchers passedcase_16: fail — committed_regex: pattern '1\\.9\\s*[-‐-―]\\s*2\\.0' against '2.0 percent'case_17: fail — committed_value: committed 3.875, expected 3.75case_18: fail — not_derivable: the committed answer does not say why the figure cannot identify a participantcase_19: fail — not_derivable: the committed answer does not say why the figure cannot identify a participantcase_20: pass — all matchers passedcase_21: fail — committed_value: committed answer carries no number: 'Not stated in the document'case_22: fail — committed_value: committed answer carries no number: 'Not stated in the document'case_23: fail — committed_regex: pattern '3\\.38\\s*[-‐-―]\\s*3\\.62' against '3.25–3.5 percent'3/23 passed | score 0.130 · 3/23 cases passed · 0 skipped | 117.00s | $0.393 | 3h ago |