Replace five near-duplicate documents with five new document types

.gitattributes

@@ -160,3 +160,8 @@ docs/plan_reading/tower_coverage_credit_01.pdf filter=lfs diff=lfs merge=lfs -te
 docs/plan_reading/water_loss_audit_01.pdf filter=lfs diff=lfs merge=lfs -text
 docs/plan_reading/water_loss_audit_02.pdf filter=lfs diff=lfs merge=lfs -text
 document_conventions.jsonl filter=lfs diff=lfs merge=lfs -text
+docs/chart_reading/borehole_core_log_01.pdf filter=lfs diff=lfs merge=lfs -text
+docs/chart_reading/gantt_schedule_board_01.pdf filter=lfs diff=lfs merge=lfs -text
+docs/plan_reading/audit_findings_report_01.pdf filter=lfs diff=lfs merge=lfs -text
+docs/plan_reading/capital_call_notice_01.pdf filter=lfs diff=lfs merge=lfs -text
+docs/plan_reading/tournament_results_pack_01.pdf filter=lfs diff=lfs merge=lfs -text

README.md

@@ -51,7 +51,7 @@ tags:
 
 ExtractBench Pro is the hard companion to [ExtractBench](https://huggingface.co/datasets/llamaindex/ExtractBench). The task is the same: given a full document and a user-written schema, return a schema-valid JSON object with the correct, exhaustive values and evidence for each one. The difference is what makes a value hard. In ExtractBench Pro, many values cannot be copied from the page. The reader has to read a curve or a drawing, apply the counting rules the issuer states once, or follow a convention set elsewhere in the document.
 
-ExtractBench Pro contains 100 documents (2,676 pages, 1–83 pages each) across 81 document types, with 143,735 expected values. Every document is synthetic: its values are fictional and fixed before the page is rendered, so the expected output is exact by construction.
+ExtractBench Pro contains 100 documents (2,622 pages, 1–83 pages each) across 86 document types, with 144,141 expected values. Every document is synthetic: its values are fictional and fixed before the page is rendered, so the expected output is exact by construction.
 
 ## Challenge subsets
 
@@ -59,18 +59,18 @@ Each document belongs to exactly one subset: the challenge a naive reader most o
 
 | Subset | Challenge | Documents | Pages | Expected values |
 |---|---|---:|---:|---:|
-| `chart_reading` | Read values printed only as a curve, trace, log track, grid cell or lookup chart, and combine them with the rules and tables of the document | 20 | 332 | 3,848 |
-| `plan_reading` | Read relations and areas that exist only as drawn geometry or topology: plats, site plans, one-line diagrams, risers, structure charts and family trees | 27 | 535 | 2,637 |
-| `counting_rules` | Count, total and classify records under the rules and definitions the issuer states once, in its own terms | 33 | 1,012 | 31,211 |
+| `chart_reading` | Read values printed only as a curve, trace, log track, schedule bar, grid cell or lookup chart, and combine them with the rules and tables of the document | 22 | 335 | 4,336 |
+| `plan_reading` | Read relations and areas that exist only as drawn geometry or topology: plats, site plans, one-line diagrams, risers, structure and org charts, family trees and tournament brackets | 30 | 581 | 2,867 |
+| `counting_rules` | Count, total and classify records under the rules and definitions the issuer states once, in its own terms | 28 | 909 | 30,899 |
 | `document_conventions` | Return the value a convention stated elsewhere sets: codes, hand-filled forms, system printouts, amendments and restatements | 20 | 797 | 106,039 |
-| **Total** | | **100** | **2,676** | **143,735** |
+| **Total** | | **100** | **2,622** | **144,141** |
 
 <details>
   <summary>Document types per subset</summary>
 
-- **Chart reading:** wireline logs (3), time-location charts (2), agency staffing invoice, aggregate QC report, chemotherapy verification, contractor dispute binder, dam operations report, drilling report bundle, earthwork quantity packet, integrity filing annex, pipeline inspection report, pump station calculation, QT audit bulletin, stock declaration file, titration packet, track chart, well completion filing
-- **Plan reading:** credit facility binders (2), fund structure files (2), HVAC balancing reports (2), insurance programme files (2), network change requests (2), one-line commissioning binders (2), water loss audits (2), assessment billing run, capacity change pack, claims triage bundle, eligibility appeals file, fibre splice record, medical gas shutdown notice, parking variance hearing, probate court file, procurement audit pack, rezoning notice packet, right-of-way acquisition packet, site plan review, tower coverage credit
-- **Counting rules:** library aid claims (3), election canvass binders (2), transit ridership reviews (2), travel expense binders (2), care home census review, circulation audit, clinical listing, college cohort file, container demurrage claim, demand response settlement, engine utilisation return, export credit policy results, franchise royalty report, hotel room block review, import duty reconciliation, joint interest billing review, lab safety review, landfill diversion return, mortgage servicing report, payroll audit packet, premium audit worksheet, purchase contract review, school enrollment review, show entry check, spread betting account, subscription seat review, TV post-buy affidavit, vendor statement reconciliation
+- **Chart reading:** wireline logs (3), time-location charts (2), agency staffing invoice, aggregate QC report, borehole core log, chemotherapy verification, contractor dispute binder, dam operations report, drilling report bundle, earthwork quantity packet, Gantt schedule board, integrity filing annex, pipeline inspection report, pump station calculation, QT audit bulletin, stock declaration file, titration packet, track chart, well completion filing
+- **Plan reading:** credit facility binders (2), fund structure files (2), HVAC balancing reports (2), insurance programme files (2), network change requests (2), one-line commissioning binders (2), water loss audits (2), assessment billing run, audit findings report, capacity change pack, capital call notice, claims triage bundle, eligibility appeals file, fibre splice record, medical gas shutdown notice, parking variance hearing, probate court file, procurement audit pack, rezoning notice packet, right-of-way acquisition packet, site plan review, tournament results pack, tower coverage credit
+- **Counting rules:** care home census review, circulation audit, clinical listing, college cohort file, container demurrage claim, demand response settlement, election canvass binder, engine utilisation return, export credit policy results, franchise royalty report, hotel room block review, import duty reconciliation, joint interest billing review, lab safety review, landfill diversion return, library aid claim, mortgage servicing report, payroll audit packet, premium audit worksheet, purchase contract review, school enrollment review, show entry check, spread betting account, subscription seat review, transit ridership review, travel expense binder, TV post-buy affidavit, vendor statement reconciliation
 - **Document conventions:** GL trial balances (2), loss runs (2), regulations (2), utility invoices (2), benefits census, customs entry (CBP 7501), grant award form, insurance schedule, investor report, mortgage application (URLA 1003), patient record, payment remittance, property bulletin, SEC 13F report, UB-04 claim form, UCC search form
 
 </details>
@@ -81,16 +81,16 @@ Each document is also tagged by how most of its pages were captured, by length a
 
 | Perception | Documents | What most pages look like |
 |---|---:|---|
-| `digital` | 21 | Born-digital pages with a text layer |
+| `digital` | 25 | Born-digital pages with a text layer |
 | `image` | 8 | Clean page images without scanner noise |
-| `scan` | 45 | Light or medium scan |
-| `heavy_scan` | 19 | Heavy or extreme scan |
+| `scan` | 44 | Light or medium scan |
+| `heavy_scan` | 16 | Heavy or extreme scan |
 | `handwritten` | 7 | Hand-filled forms or handwritten values |
 
-- **Length:** `short` (10 pages or fewer) 6 documents, `medium` (11–50 pages) 83 documents, `long` (more than 50 pages) 11 documents.
-- **Domain:** `D1` finance (26), `D2` energy (18), `D3` government (19), `D5` supply chain (8), `D6` healthcare (13), `D7` legal (3), `D8` real estate (13), using the ExtractBench domain codes.
+- **Length:** `short` (10 pages or fewer) 8 documents, `medium` (11–50 pages) 81 documents, `long` (more than 50 pages) 11 documents.
+- **Domain:** `D1` finance (27), `D2` energy (19), `D3` government (16), `D5` supply chain (9), `D6` healthcare (13), `D7` legal (3), `D8` real estate (13), using the ExtractBench domain codes.
 
-The six short documents are single plotter sheets 13 to 17 feet long (well logs, a track chart and time-location charts), scanned or photocopied, with no text layer; each asks for 446 to 715 values. Charts and drawings in other documents are often page images too, so their values cannot be read from a text layer. Some scanned documents carry a text layer that does not match the printed page. The printed page is authoritative. One document prints two pages on each sheet, turned sideways.
+Seven of the eight short documents are single plotter sheets 13 to 17 feet long (well logs, a borehole core log, a track chart and time-location charts), scanned or photocopied, with no text layer; each asks for 390 to 715 values. The eighth is a two-page Gantt schedule board. Charts and drawings in other documents are often page images too, so their values cannot be read from a text layer. Some scanned documents carry a text layer that does not match the printed page. The printed page is authoritative. One document prints two pages on each sheet, turned sideways.
 
 ## Usage
 
@@ -139,9 +139,9 @@ Scoring follows ExtractBench and is fully deterministic:
 
 The evaluation dataset files are:
 
-- [chart_reading.jsonl](chart_reading.jsonl): 20 test cases across 332 pages
-- [plan_reading.jsonl](plan_reading.jsonl): 27 test cases across 535 pages
-- [counting_rules.jsonl](counting_rules.jsonl): 33 test cases across 1,012 pages
+- [chart_reading.jsonl](chart_reading.jsonl): 22 test cases across 335 pages
+- [plan_reading.jsonl](plan_reading.jsonl): 30 test cases across 581 pages
+- [counting_rules.jsonl](counting_rules.jsonl): 28 test cases across 909 pages
 - [document_conventions.jsonl](document_conventions.jsonl): 20 test cases across 797 pages
 - [docs/](https://huggingface.co/datasets/llamaindex/ExtractBenchPro/tree/main/docs): source PDFs organized by subset
 

counting_rules.jsonl

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:123d2a5126261eec99e4144f3e13296be2acfc0b145e47de7c641f69c8758cf5
-size 25845809
+oid sha256:675f37d1ad68027a4c1944c3b439de76d1ee0882b3f20bb33f5a278073df5173
+size 24812634

docs/{counting_rules/election_canvass_binder_02.pdf → chart_reading/borehole_core_log_01.pdf}

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d964cb1ea9f547f6974a7527d6a4caf958a6ceb64c21c6b23ed88ba37d3f43d2
-size 8174185
+oid sha256:b581086be45dcef5968fba69634d832232510567338ce613a290d1b797dda4f9
+size 6541172

docs/{counting_rules/travel_expense_binder_02.pdf → chart_reading/gantt_schedule_board_01.pdf}

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f825c3da81e85116282872f6b14caeb2d4b77ba3324392ae8e7362c7e347efe3
-size 5781262
+oid sha256:13cde5942365f245f91476d725959d974bf18f23d43f4d78476df0ce59c42f1c
+size 555353

docs/counting_rules/library_aid_claim_01.pdf

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fbe2d7755565f0ad74c1e737854fbd61a815df267387e948205114a697ea31ae
-size 13464216
+oid sha256:c40e91649a98e233b4fcab0fccc05a6898061334a48b627e88f40fd914a09b51
+size 13154061

docs/counting_rules/library_aid_claim_02.pdf

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:c40e91649a98e233b4fcab0fccc05a6898061334a48b627e88f40fd914a09b51
-size 13154061

docs/counting_rules/transit_ridership_review_01.pdf

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:837096e871a497fe820d293cd854df651378241b8a72647da16f8ab0000adda9
-size 12419217
+oid sha256:0937a5979e36519f1334c27ff4d6297fac698999e219a45da460fe8b75be6627
+size 12421497

docs/{counting_rules/library_aid_claim_03.pdf → plan_reading/audit_findings_report_01.pdf}

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:796c931071975752108f6acd2f93986a9b7c8bd844f7f0459e060d39210cbc1f
-size 13056365
+oid sha256:7f3525058bd6e1e0fc8d8f617f5fbc2a905afdb1c4c758d5226b45af13d69c4b
+size 105291

docs/{counting_rules/transit_ridership_review_02.pdf → plan_reading/capital_call_notice_01.pdf}

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0937a5979e36519f1334c27ff4d6297fac698999e219a45da460fe8b75be6627
-size 12421497
+oid sha256:ac63574bb72c376c0c5bc1ccb2af0398fed69a8579edb088044df7ca49f10475
+size 250896

docs/plan_reading/tournament_results_pack_01.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c6d4fb3ca2023d338ec2706e6f97690f9b2dfb144edc4ef1ddfb0d67575170fb
+size 860051