{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "Tamberly Freight runs 260 refrigerated trucks out of three depots in the English Midlands.", "subjects": ["Tamberly Freight"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "Until July 2026 our route planner ran once a night, at 02:00, and drivers received the next day's routes on a printed manifest.", "subjects": ["Tamberly Freight"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "In July we replaced the nightly batch with an event-driven planner that re-plans a route whenever an order is added, cancelled or delayed.", "subjects": ["Tamberly Freight"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-001", "run": 1, "outcome": "spurious", "stated": 0.95, "content": "Each change is published to a message queue, and the planner recomputes only the routes that touch the affected depot.", "subjects": ["Tamberly Freight"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "The results after eight weeks: the average time from an order being placed to it appearing on a driver's route fell from 19 hours to 40 minutes.", "subjects": ["Tamberly Freight"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "Empty running, the share of kilometres driven with no load, dropped from 23 percent to 17 percent.", "subjects": ["Tamberly Freight"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "We considered buying a commercial routing service instead.", "subjects": ["Tamberly Freight"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "We rejected it because none of the three vendors we spoke to supported temperature-zone constraints, and every one of our loads has one.", "subjects": ["Tamberly Freight"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "The migration took a team of five engineers eleven weeks.", "subjects": ["Tamberly Freight"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "Printed manifests are still produced as a backup and will be retired in January 2027.", "subjects": ["Tamberly Freight"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 1, "outcome": "matched", "stated": 0.95, "content": "Ashcombe Vale's public library reopened on Saturday in the town's former bus depot on Quarry Road on 20 September 2026.", "subjects": ["Ashcombe Vale's public library", "former bus depot on Quarry Road"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 1, "outcome": "spurious", "stated": 0.95, "content": "The old library on the High Street closed on 20 September 2026 because of subsidence.", "subjects": ["old library on the High Street"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 1, "outcome": "matched", "stated": 0.95, "content": "The conversion of the former bus depot into a library cost \u00a33.4 million.", "subjects": ["conversion of the former bus depot into a library"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 1, "outcome": "matched", "stated": 0.95, "content": "\u00a31.1 million of the conversion cost came from a national lottery heritage grant.", "subjects": ["conversion of the former bus depot into a library"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 1, "outcome": "spurious", "stated": 0.95, "content": "The district council provided the remaining \u00a32.3 million for the conversion of the former bus depot into a library.", "subjects": ["district council"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 1, "outcome": "matched", "stated": 0.95, "content": "The new library has 1,850 square metres of floor space.", "subjects": ["new library"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 1, "outcome": "matched", "stated": 0.95, "content": "The new library has about twice the floor space of the old library on the High Street.", "subjects": ["new library", "old library on the High Street"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 1, "outcome": "matched", "stated": 0.95, "content": "The new library holds 46,000 books.", "subjects": ["new library"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 1, "outcome": "matched", "stated": 0.95, "content": "The new library is open seven days a week, from 9:00 to 20:00 on weekdays and 10:00 to 16:00 at weekends.", "subjects": ["new library"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 1, "outcome": "spurious", "stated": 0.95, "content": "The old library on the High Street was closed on Sundays.", "subjects": ["old library on the High Street"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 1, "outcome": "matched", "stated": 0.95, "content": "The former depot's inspection pit has been kept under a glass floor in the children's section of the new library.", "subjects": ["former depot's inspection pit", "new library"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 1, "outcome": "matched", "stated": 0.95, "content": "Head librarian Ines Calloway said, 'We wanted people to see what the building used to be.'", "subjects": ["Ines Calloway", "building"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 1, "outcome": "matched", "stated": 0.85, "content": "Around 1,200 people visited the new library on its opening day, which was 20 September 2026.", "subjects": ["new library"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 1, "outcome": "matched", "stated": 0.85, "content": "A caf\u00e9 run by the Vale Community Kitchen charity will open in the new library in November.", "subjects": ["caf\u00e9 run by the Vale Community Kitchen charity", "new library"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-essay-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "Hearthwell Bakery was founded in 1994 as a cooperative by six bakers who had all been laid off when the Morrow & Pike flour mill in Lindfield closed.", "subjects": ["Hearthwell Bakery", "Morrow & Pike flour mill", "Lindfield"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-essay-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "Each member put in \u00a32,000, and the cooperative still pays every member the same hourly wage.", "subjects": ["Hearthwell Bakery"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-essay-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "Crumbworth, which opened on the same street in 1996, now supplies 140 supermarkets from a factory outside the town.", "subjects": ["Crumbworth", "Hearthwell Bakery"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-essay-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "Hearthwell has one shop and one oven, a wood-fired oven built in 1994 that is still in daily use.", "subjects": ["Hearthwell Bakery"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-essay-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "The cooperative has twice voted against expansion: in 2008, when a supermarket chain offered a supply contract, and in 2019, when the council offered it a second unit at a reduced rent. Both votes were unanimous.", "subjects": ["Hearthwell Bakery"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-essay-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "Membership has grown only from six to eleven in thirty-two years.", "subjects": ["Hearthwell Bakery"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-essay-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "The shop sells around 900 loaves on a Saturday and closes when they are gone.", "subjects": ["Hearthwell Bakery"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-essay-001", "run": 1, "outcome": "spurious", "stated": 0.75, "content": "I think Hearthwell's refusal to grow is the reason it has survived; a bakery that cannot be bought cheaply is a bakery nobody tries to buy.", "subjects": ["Hearthwell Bakery"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-interview-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "The Corran Point station measures sea level and water temperature at the mouth of the Corran estuary every ten minutes.", "subjects": ["Corran Point station"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-interview-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "The Corran Point station has run without a gap since 1987.", "subjects": ["Corran Point station"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-interview-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "The Corran Point station is the longest unbroken record on this coast.", "subjects": ["Corran Point station"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-interview-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "Mean sea level at the Corran Point station is 11 centimetres higher than it was in 1987.", "subjects": ["Corran Point station"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-interview-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "Summer surface temperature at the Corran Point station has risen by 1.4 degrees since 1987.", "subjects": ["Corran Point station"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-interview-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "The Corran Estuary Trust owns the Corran Point station.", "subjects": ["Corran Estuary Trust", "Corran Point station"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-interview-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "Kofi Ansah-Brandt is the only full-time staff member of the Corran Point station.", "subjects": ["Kofi Ansah-Brandt", "Corran Point station"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-interview-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "Two volunteers check the sensors at the Corran Point station each week.", "subjects": ["two volunteers", "Corran Point station"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-interview-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "A salinity sensor will be added to the Corran Point station in October.", "subjects": ["Corran Point station"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-interview-001", "run": 1, "outcome": "matched", "stated": 0.3, "content": "The salt wedge is moving further upstream each year, according to Kofi Ansah-Brandt's guess.", "subjects": ["salt wedge"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "Skerrin Island Ferries carried 412,000 passengers in the six months to 30 June 2026.", "subjects": ["Skerrin Island Ferries"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "Vehicle crossings rose from 58,000 to 63,500.", "subjects": ["Skerrin Island Ferries"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "Revenue for the half was \u00a39.8 million.", "subjects": ["Skerrin Island Ferries"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "Operating profit for the half was \u00a31.2 million.", "subjects": ["Skerrin Island Ferries"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "Fuel was the largest single cost at \u00a32.6 million.", "subjects": ["Skerrin Island Ferries"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "The new hybrid vessel MV Aird Mhor entered service on 3 March.", "subjects": ["MV Aird Mhor"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "MV Aird Mhor now makes eight of the eighteen daily crossings.", "subjects": ["MV Aird Mhor"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "MV Aird Mhor uses 31 percent less diesel than the vessel it replaced.", "subjects": ["MV Aird Mhor"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "Punctuality, measured as departures within five minutes of the timetable, was 94.2 percent.", "subjects": ["Skerrin Island Ferries"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "Forty-one sailings were cancelled for weather in the first half of 2026.", "subjects": ["Skerrin Island Ferries"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "Sixty-six sailings were cancelled for weather in the first half of 2025.", "subjects": ["Skerrin Island Ferries"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "The board has approved an order for a second hybrid vessel, due in 2028.", "subjects": ["Skerrin Island Ferries"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-incident-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "Card payments through Quillon failed for 47 minutes on 4 September, from 14:12 to 14:59 UTC.", "subjects": ["Quillon"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-incident-001", "run": 1, "outcome": "matched", "stated": 0.75, "content": "Roughly 18,000 transactions were declined; the exact figure will change as merchants report retries.", "subjects": ["Quillon"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-incident-001", "run": 1, "outcome": "matched", "stated": 0.75, "content": "Our leading hypothesis is that a certificate rotation on the card-network link at 14:10 left one of three gateway nodes presenting an expired certificate.", "subjects": ["Quillon"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-incident-001", "run": 1, "outcome": "spurious", "stated": 0.6, "content": "We think this is likely but have not reproduced it.", "subjects": ["Quillon"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-incident-001", "run": 1, "outcome": "spurious", "stated": 0.9, "content": "We have ruled out a database fault: the payments database showed normal latency throughout.", "subjects": ["Quillon"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-incident-001", "run": 1, "outcome": "matched", "stated": 0.6, "content": "It is unclear why traffic was not drained from the affected node automatically.", "subjects": ["Quillon"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-incident-001", "run": 1, "outcome": "matched", "stated": 0.6, "content": "The health check may have been testing the wrong endpoint, but we have not confirmed that.", "subjects": ["Quillon"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-incident-001", "run": 1, "outcome": "spurious", "stated": 0.9, "content": "A full review will be published by 30 September.", "subjects": ["Quillon"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-incident-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "No customer card data was exposed.", "subjects": ["Quillon"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-trial-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "Nine member farms undersowed winter wheat with white clover in 2025 and 2026.", "subjects": ["Brackenridge Farm Network"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-trial-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "Each of the nine farms left a matched strip on each farm without clover.", "subjects": ["Brackenridge Farm Network"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-trial-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "A tenth farm dropped out after its strip flooded in February.", "subjects": ["Brackenridge Farm Network"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-trial-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "Across the nine farms, nitrogen fertiliser use on the clover strips was 28 percent lower.", "subjects": ["Brackenridge Farm Network"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-trial-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "Yields were between 4 percent lower and 2 percent higher than the matched strips.", "subjects": ["Brackenridge Farm Network"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-trial-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "Earthworm counts were higher on the clover strips on seven of the nine farms.", "subjects": ["Brackenridge Farm Network"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-trial-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "A third year is planned if funding from the regional soils fund is renewed in December.", "subjects": ["Brackenridge Farm Network"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-trial-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "Our best estimate is that a farm adopting undersowing would break even within three seasons.", "subjects": ["Brackenridge Farm Network"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-trial-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "That estimate rests on this year's fertiliser price and could easily be wrong.", "subjects": ["Brackenridge Farm Network"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "MS 1172 is a two-page letter describing the collapse of the Wheal Tressa mine engine house.", "subjects": ["MS 1172"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "MS 1172 is unsigned.", "subjects": ["MS 1172"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "MS 1172 was given to the Record Office in 1964 by a private donor, and nothing is known of where it was before that.", "subjects": ["MS 1172"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 1, "outcome": "matched", "stated": 0.85, "content": "The letter has traditionally been attributed to the mine captain Josiah Penhallow.", "subjects": ["MS 1172"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 1, "outcome": "matched", "stated": 0.85, "content": "The handwriting is close to that of his surviving account books.", "subjects": ["Josiah Penhallow"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 1, "outcome": "matched", "stated": 0.85, "content": "The writer refers to 'my men'.", "subjects": ["MS 1172"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 1, "outcome": "spurious", "stated": 0.85, "content": "Dr Lowenna Tresize has argued instead that it was written by his daughter, Keziah, who kept the mine's correspondence.", "subjects": ["Keziah Penhallow"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 1, "outcome": "matched", "stated": 0.85, "content": "She points out that the letter uses spellings found in Keziah's letters and never in Josiah's.", "subjects": ["MS 1172"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 1, "outcome": "spurious", "stated": 0.95, "content": "The Record Office does not take a view.", "subjects": ["Dunmere County Record Office"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "We catalogue the letter as 'attributed to Josiah Penhallow, attribution disputed'.", "subjects": ["Dunmere County Record Office"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "The paper's watermark dates it to between 1838 and 1846.", "subjects": ["MS 1172"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 1, "outcome": "spurious", "stated": 0.85, "content": "The collapse it describes is usually dated to 1841, but no newspaper report of it has been found.", "subjects": ["The collapse described in MS 1172"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-forecast-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "The council's spending this year, 2026\u201327, is \u00a314.2 million.", "subjects": ["Ferrisford Town Council"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-forecast-001", "run": 1, "outcome": "matched", "stated": 0.75, "content": "For 2027\u201328 we forecast a gap of between \u00a30.9 million and \u00a31.6 million between expected spending and expected income.", "subjects": ["Ferrisford Town Council"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-forecast-001", "run": 1, "outcome": "spurious", "stated": 0.65, "content": "The width of that range is mostly the leisure centre.", "subjects": ["Ferrisford Town Council", "Ferrisford leisure centre"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-forecast-001", "run": 1, "outcome": "matched", "stated": 0.75, "content": "If councillors vote in November to close the Ferrisford pool, the gap would fall by about \u00a30.5 million a year.", "subjects": ["Ferrisford Town Council", "Ferrisford pool"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-forecast-001", "run": 1, "outcome": "matched", "stated": 0.75, "content": "If they keep it open, the pool's boiler is expected to need replacing within two years, at an estimated cost of \u00a30.8 million.", "subjects": ["Ferrisford Town Council", "Ferrisford pool"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-forecast-001", "run": 1, "outcome": "matched", "stated": 0.75, "content": "We have assumed council tax rises by the maximum 4.99 percent allowed without a local referendum.", "subjects": ["Ferrisford Town Council"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-forecast-001", "run": 1, "outcome": "spurious", "stated": 0.75, "content": "Each percentage point below that adds roughly \u00a385,000 to the gap.", "subjects": ["Ferrisford Town Council"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-forecast-001", "run": 1, "outcome": "matched", "stated": 0.75, "content": "Our income forecast assumes car-parking revenue returns to its 2019 level.", "subjects": ["Ferrisford Town Council"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-forecast-001", "run": 1, "outcome": "spurious", "stated": 0.65, "content": "We think that is optimistic; it has recovered to about 80 percent of that level so far.", "subjects": ["Ferrisford Town Council"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-forecast-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "A draft budget will go to full council on 9 February 2027.", "subjects": ["Ferrisford Town Council"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-correction-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "The contract value for the repairs to Calderhaven's north harbour wall is \u00a31.7 million.", "subjects": ["Calderhaven's north harbour wall repair contract"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-correction-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "The original estimate for the repairs to Calderhaven's north harbour wall was \u00a32.1 million.", "subjects": ["Calderhaven's north harbour wall repair contract"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-correction-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "Repairs to Calderhaven's north harbour wall will not start until spring 2027.", "subjects": ["Calderhaven's north harbour wall repairs"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-correction-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "Work on the repairs to Calderhaven's north harbour wall had been due to start this month (September 2026).", "subjects": ["Calderhaven's north harbour wall repairs"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-correction-001", "run": 1, "outcome": "matched", "stated": 0.75, "content": "The delay in repairs to Calderhaven's north harbour wall is blamed on a shortage of the granite needed to match the original wall.", "subjects": ["Calderhaven's north harbour wall repairs"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-correction-001", "run": 1, "outcome": "spurious", "stated": 0.75, "content": "The contractor, Pendry Marine, claims the delay in repairs to Calderhaven's north harbour wall is due to the authority not yet approving the final design.", "subjects": ["Pendry Marine", "Calderhaven's north harbour wall repairs"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-correction-001", "run": 1, "outcome": "matched", "stated": 0.75, "content": "Pendry Marine said that granite is available for the repairs to Calderhaven's north harbour wall.", "subjects": ["Pendry Marine", "Calderhaven's north harbour wall repairs"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-correction-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "Fishing boats have been unable to use the north berths since January 2026.", "subjects": ["Calderhaven's north berths"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-correction-001", "run": 1, "outcome": "spurious", "stated": 0.75, "content": "The closure of the north berths in Calderhaven has cost the fishing fleet somewhere around \u00a3300,000 in lost landings.", "subjects": ["Calderhaven's north berths closure"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-correction-001", "run": 1, "outcome": "matched", "stated": 0.75, "content": "The harbourmaster estimates that the closure of the north berths in Calderhaven has cost the fishing fleet somewhere around \u00a3300,000 in lost landings.", "subjects": ["Calderhaven's north berths closure", "harbourmaster"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-correction-001", "run": 1, "outcome": "matched", "stated": 0.95, "content": "A public meeting will be held at the harbour office on 2 October 2026.", "subjects": ["public meeting", "Calderhaven harbour office"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "Tamberly Freight runs 260 refrigerated trucks out of three depots in the English Midlands.", "subjects": ["Tamberly Freight"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "Until July 2026 our route planner ran once a night, at 02:00, and drivers received the next day's routes on a printed manifest.", "subjects": ["Tamberly Freight", "drivers"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "In July we replaced the nightly batch with an event-driven planner that re-plans a route whenever an order is added, cancelled or delayed.", "subjects": ["Tamberly Freight"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-001", "run": 2, "outcome": "spurious", "stated": 0.95, "content": "Each change is published to a message queue, and the planner recomputes only the routes that touch the affected depot.", "subjects": ["Tamberly Freight"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "The results after eight weeks: the average time from an order being placed to it appearing on a driver's route fell from 19 hours to 40 minutes.", "subjects": ["Tamberly Freight"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "Empty running, the share of kilometres driven with no load, dropped from 23 percent to 17 percent.", "subjects": ["Tamberly Freight"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "We considered buying a commercial routing service instead.", "subjects": ["Tamberly Freight"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "We rejected it because none of the three vendors we spoke to supported temperature-zone constraints, and every one of our loads has one.", "subjects": ["Tamberly Freight"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "The migration took a team of five engineers eleven weeks.", "subjects": ["Tamberly Freight"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "Printed manifests are still produced as a backup and will be retired in January 2027.", "subjects": ["Tamberly Freight"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 2, "outcome": "matched", "stated": 0.95, "content": "Ashcombe Vale's public library reopened on Saturday in the town's former bus depot on Quarry Road.", "subjects": ["Ashcombe Vale's public library", "former bus depot on Quarry Road"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 2, "outcome": "matched", "stated": 0.95, "content": "The old building on the High Street closed because of subsidence two years before the library reopened.", "subjects": ["old building on the High Street"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 2, "outcome": "matched", "stated": 0.95, "content": "The conversion of the former bus depot into a library cost \u00a33.4 million.", "subjects": ["conversion of the former bus depot into a library"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 2, "outcome": "matched", "stated": 0.95, "content": "\u00a31.1 million of the conversion cost came from a national lottery heritage grant.", "subjects": ["conversion of the former bus depot into a library"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 2, "outcome": "matched", "stated": 0.95, "content": "The rest of the conversion cost came from the district council.", "subjects": ["conversion of the former bus depot into a library"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 2, "outcome": "matched", "stated": 0.95, "content": "The new library has 1,850 square metres of floor space.", "subjects": ["new library"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 2, "outcome": "matched", "stated": 0.85, "content": "The new library has about twice the floor space of the old building.", "subjects": ["new library", "old building"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 2, "outcome": "matched", "stated": 0.95, "content": "The new library holds 46,000 books.", "subjects": ["new library"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 2, "outcome": "matched", "stated": 0.95, "content": "The new library is open seven days a week, from 9:00 to 20:00 on weekdays and 10:00 to 16:00 at weekends.", "subjects": ["new library"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 2, "outcome": "matched", "stated": 0.95, "content": "The old library closed on Sundays.", "subjects": ["old library"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 2, "outcome": "matched", "stated": 0.95, "content": "The former depot's inspection pit has been kept under a glass floor in the children's section.", "subjects": ["former depot's inspection pit"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 2, "outcome": "matched", "stated": 0.95, "content": "Ines Calloway, head librarian, said: 'We wanted people to see what the building used to be.'", "subjects": ["Ines Calloway"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 2, "outcome": "spurious", "stated": 0.95, "content": "Around 1,200 people visited on the opening day.", "subjects": ["Ashcombe Vale's public library"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 2, "outcome": "matched", "stated": 0.95, "content": "A caf\u00e9 run by the Vale Community Kitchen charity will open in the building in November.", "subjects": ["caf\u00e9 run by the Vale Community Kitchen charity", "building"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 2, "outcome": "spurious", "stated": 0.95, "content": "The article was published on 20 September 2026.", "subjects": []}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-essay-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "Hearthwell Bakery was founded in 1994 as a cooperative by six bakers who had all been laid off when the Morrow & Pike flour mill in Lindfield closed.", "subjects": ["Hearthwell Bakery", "Morrow & Pike flour mill", "Lindfield"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-essay-001", "run": 2, "outcome": "matched", "stated": 0.85, "content": "Each member put in \u00a32,000, and the cooperative still pays every member the same hourly wage.", "subjects": ["Hearthwell Bakery"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-essay-001", "run": 2, "outcome": "matched", "stated": 0.92, "content": "Its rivals grew. Crumbworth, which opened on the same street in 1996, now supplies 140 supermarkets from a factory outside the town.", "subjects": ["Crumbworth", "Hearthwell Bakery"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-essay-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "Hearthwell has one shop and one oven, a wood-fired oven built in 1994 that is still in daily use.", "subjects": ["Hearthwell Bakery"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-essay-001", "run": 2, "outcome": "matched", "stated": 0.92, "content": "The cooperative has twice voted against expansion: in 2008, when a supermarket chain offered a supply contract, and in 2019, when the council offered it a second unit at a reduced rent. Both votes were unanimous.", "subjects": ["Hearthwell Bakery"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-essay-001", "run": 2, "outcome": "matched", "stated": 0.9, "content": "Membership has grown only from six to eleven in thirty-two years.", "subjects": ["Hearthwell Bakery"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-essay-001", "run": 2, "outcome": "matched", "stated": 0.8, "content": "The shop sells around 900 loaves on a Saturday and closes when they are gone.", "subjects": ["Hearthwell Bakery"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-essay-001", "run": 2, "outcome": "spurious", "stated": 0.5, "content": "I think Hearthwell's refusal to grow is the reason it has survived; a bakery that cannot be bought cheaply is a bakery nobody tries to buy.", "subjects": ["Hearthwell Bakery"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-interview-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "The Corran Point station measures sea level and water temperature at the mouth of the Corran estuary every ten minutes.", "subjects": ["Corran Point station"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-interview-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "The Corran Point station has run without a gap since 1987.", "subjects": ["Corran Point station"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-interview-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "The Corran Point station is the longest unbroken record on this coast.", "subjects": ["Corran Point station"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-interview-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "Mean sea level at the Corran Point station is 11 centimetres higher than it was in 1987.", "subjects": ["Corran Point station"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-interview-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "Summer surface temperature at the Corran Point station has risen by 1.4 degrees since 1987.", "subjects": ["Corran Point station"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-interview-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "The Corran Estuary Trust owns the Corran Point station.", "subjects": ["Corran Estuary Trust", "Corran Point station"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-interview-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "Kofi Ansah-Brandt is the only full-time staff member of the Corran Point station.", "subjects": ["Kofi Ansah-Brandt", "Corran Point station"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-interview-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "Two volunteers check the sensors at the Corran Point station each week.", "subjects": ["two volunteers", "Corran Point station"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-interview-001", "run": 2, "outcome": "matched", "stated": 0.75, "content": "A salinity sensor will be added to the Corran Point station in October.", "subjects": ["Corran Point station"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-interview-001", "run": 2, "outcome": "matched", "stated": 0.4, "content": "Kofi Ansah-Brandt guesses that the salt wedge is moving further upstream each year at the Corran Point station.", "subjects": ["Kofi Ansah-Brandt", "Corran Point station"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-interview-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "There is no data to show that the salt wedge is moving further upstream each year at the Corran Point station.", "subjects": ["Corran Point station"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "Skerrin Island Ferries carried 412,000 passengers in the six months to 30 June 2026.", "subjects": ["Skerrin Island Ferries"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "Skerrin Island Ferries carried 371,000 passengers in the six months to 30 June 2025.", "subjects": ["Skerrin Island Ferries"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "Vehicle crossings rose from 58,000 to 63,500.", "subjects": ["Skerrin Island Ferries"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "Revenue for the half was \u00a39.8 million.", "subjects": ["Skerrin Island Ferries"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "Operating profit for the half was \u00a31.2 million.", "subjects": ["Skerrin Island Ferries"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "Fuel was the largest single cost at \u00a32.6 million.", "subjects": ["Skerrin Island Ferries"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "The new hybrid vessel MV Aird Mhor entered service on 3 March 2026.", "subjects": ["MV Aird Mhor"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "The new hybrid vessel MV Aird Mhor now makes eight of the eighteen daily crossings.", "subjects": ["MV Aird Mhor"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "The new hybrid vessel MV Aird Mhor uses 31 percent less diesel than the vessel it replaced.", "subjects": ["MV Aird Mhor"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "Punctuality, measured as departures within five minutes of the timetable, was 94.2 percent.", "subjects": ["Skerrin Island Ferries"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "Forty-one sailings were cancelled for weather in the first half of 2026.", "subjects": ["Skerrin Island Ferries"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "Sixty-six sailings were cancelled for weather in the first half of 2025.", "subjects": ["Skerrin Island Ferries"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "The board has approved an order for a second hybrid vessel, due in 2028.", "subjects": ["Skerrin Island Ferries"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-incident-001", "run": 2, "outcome": "matched", "stated": 0.92, "content": "Card payments through Quillon failed for 47 minutes on 4 September, from 14:12 to 14:59 UTC.", "subjects": ["Quillon", "card payments"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-incident-001", "run": 2, "outcome": "matched", "stated": 0.88, "content": "That window is confirmed from our gateway logs.", "subjects": ["Quillon", "gateway logs"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-incident-001", "run": 2, "outcome": "matched", "stated": 0.78, "content": "Roughly 18,000 transactions were declined; the exact figure will change as merchants report retries.", "subjects": ["Quillon", "transactions"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-incident-001", "run": 2, "outcome": "matched", "stated": 0.73, "content": "Our leading hypothesis is that a certificate rotation on the card-network link at 14:10 left one of three gateway nodes presenting an expired certificate.", "subjects": ["Quillon", "card-network link", "gateway nodes"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-incident-001", "run": 2, "outcome": "spurious", "stated": 0.72, "content": "We think this is likely but have not reproduced it.", "subjects": ["Quillon"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-incident-001", "run": 2, "outcome": "spurious", "stated": 0.91, "content": "We have ruled out a database fault: the payments database showed normal latency throughout.", "subjects": ["Quillon", "payments database"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-incident-001", "run": 2, "outcome": "matched", "stated": 0.65, "content": "It is unclear why traffic was not drained from the affected node automatically.", "subjects": ["Quillon", "affected node"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-incident-001", "run": 2, "outcome": "matched", "stated": 0.67, "content": "The health check may have been testing the wrong endpoint, but we have not confirmed that.", "subjects": ["Quillon", "health check"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-incident-001", "run": 2, "outcome": "spurious", "stated": 0.94, "content": "A full review will be published by 30 September.", "subjects": ["Quillon"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-incident-001", "run": 2, "outcome": "matched", "stated": 0.93, "content": "No customer card data was exposed.", "subjects": ["Quillon", "customer card data"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-trial-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "Nine member farms undersowed winter wheat with white clover in 2025 and 2026, leaving a matched strip on each farm without clover.", "subjects": ["Brackenridge Farm Network", "member farms"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-trial-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "A tenth farm dropped out after its strip flooded in February and is excluded from every figure here.", "subjects": ["tenth farm"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-trial-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "Across the nine farms, nitrogen fertiliser use on the clover strips was 28 percent lower.", "subjects": ["nine farms", "clover strips"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-trial-001", "run": 2, "outcome": "spurious", "stated": 0.85, "content": "Yields were between 4 percent lower and 2 percent higher than the matched strips; the range is wide because two farms harvested late in wet weather.", "subjects": ["nine farms", "matched strips", "two farms"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-trial-001", "run": 2, "outcome": "matched", "stated": 0.75, "content": "Earthworm counts were higher on the clover strips on seven of the nine farms, but the difference was not statistically significant and we would not lean on it.", "subjects": ["seven of the nine farms", "clover strips", "earthworm counts"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-trial-001", "run": 2, "outcome": "matched", "stated": 0.65, "content": "Our best estimate is that a farm adopting undersowing would break even within three seasons, though that rests on this year's fertiliser price and could easily be wrong.", "subjects": ["farm adopting undersowing"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-trial-001", "run": 2, "outcome": "matched", "stated": 0.85, "content": "A third year is planned if funding from the regional soils fund is renewed in December.", "subjects": ["third year", "regional soils fund"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "MS 1172 is a two-page letter describing the collapse of the Wheal Tressa mine engine house.", "subjects": ["MS 1172"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "MS 1172 was given to the Record Office in 1964 by a private donor.", "subjects": ["MS 1172"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 2, "outcome": "matched", "stated": 0.8, "content": "The letter has traditionally been attributed to the mine captain Josiah Penhallow.", "subjects": ["MS 1172"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 2, "outcome": "matched", "stated": 0.9, "content": "The handwriting is close to that of Josiah Penhallow's surviving account books.", "subjects": ["MS 1172"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "The writer refers to 'my men'.", "subjects": ["MS 1172"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 2, "outcome": "spurious", "stated": 0.8, "content": "Dr Lowenna Tresize has argued instead that the letter was written by Keziah Penhallow.", "subjects": ["Dr Lowenna Tresize", "MS 1172"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 2, "outcome": "matched", "stated": 0.75, "content": "Keziah Penhallow kept the mine's correspondence.", "subjects": ["Keziah Penhallow"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 2, "outcome": "matched", "stated": 0.8, "content": "Dr Lowenna Tresize points out that the letter uses spellings found in Keziah Penhallow's letters and never in Josiah Penhallow's.", "subjects": ["Dr Lowenna Tresize", "MS 1172", "Keziah Penhallow", "Josiah Penhallow"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 2, "outcome": "spurious", "stated": 0.95, "content": "The Record Office does not take a view.", "subjects": ["The Record Office"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "The letter is catalogued as 'attributed to Josiah Penhallow, attribution disputed'.", "subjects": ["MS 1172"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "The paper's watermark dates it to between 1838 and 1846.", "subjects": ["MS 1172"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 2, "outcome": "spurious", "stated": 0.75, "content": "The collapse it describes is usually dated to 1841.", "subjects": ["MS 1172"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "No newspaper report of the collapse has been found.", "subjects": ["the collapse of the Wheal Tressa mine engine house"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-forecast-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "The council's spending in 2026\u201327 is \u00a314.2 million.", "subjects": ["Ferrisford Town Council"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-forecast-001", "run": 2, "outcome": "matched", "stated": 0.9, "content": "For 2027\u201328, the council forecasts a gap of between \u00a30.9 million and \u00a31.6 million between expected spending and expected income.", "subjects": ["Ferrisford Town Council"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-forecast-001", "run": 2, "outcome": "matched", "stated": 0.85, "content": "The leisure centre is the main reason for the range in the forecast gap.", "subjects": ["Ferrisford Leisure Centre"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-forecast-001", "run": 2, "outcome": "matched", "stated": 0.75, "content": "If the Ferrisford pool is closed by councillors in November, the gap would fall by about \u00a30.5 million a year.", "subjects": ["Ferrisford Pool"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-forecast-001", "run": 2, "outcome": "matched", "stated": 0.8, "content": "If the Ferrisford pool remains open, the pool's boiler is expected to need replacing within two years at an estimated cost of \u00a30.8 million.", "subjects": ["Ferrisford Pool"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-forecast-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "The council has assumed that council tax will rise by the maximum 4.99 percent allowed without a local referendum.", "subjects": ["Ferrisford Town Council"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-forecast-001", "run": 2, "outcome": "spurious", "stated": 0.8, "content": "Each percentage point below the 4.99 percent increase adds roughly \u00a385,000 to the gap.", "subjects": ["Ferrisford Town Council"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-forecast-001", "run": 2, "outcome": "matched", "stated": 0.9, "content": "The income forecast assumes that car-parking revenue will return to its 2019 level.", "subjects": ["Ferrisford Town Council"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-forecast-001", "run": 2, "outcome": "matched", "stated": 0.8, "content": "The council thinks the assumption of car-parking revenue returning to its 2019 level is optimistic.", "subjects": ["Ferrisford Town Council"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-forecast-001", "run": 2, "outcome": "matched", "stated": 0.85, "content": "Car-parking revenue has recovered to about 80 percent of its 2019 level so far.", "subjects": ["Ferrisford Town Council"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-forecast-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "A draft budget will go to full council on 9 February 2027.", "subjects": ["Ferrisford Town Council"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-correction-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "The repair of Calderhaven's north harbour wall will not start until spring 2027.", "subjects": ["Calderhaven's north harbour wall"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-correction-001", "run": 2, "outcome": "matched", "stated": 0.75, "content": "The contractor, Pendry Marine, said that the authority had not yet approved the final design.", "subjects": ["Pendry Marine", "the harbour authority"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-correction-001", "run": 2, "outcome": "matched", "stated": 0.8, "content": "The authority blamed the delay on a shortage of the granite needed to match the original wall.", "subjects": ["the harbour authority"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-correction-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "The repair of the north harbour wall was due to start this month.", "subjects": ["Calderhaven's north harbour wall"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-correction-001", "run": 2, "outcome": "matched", "stated": 0.65, "content": "The harbourmaster estimates that the closure of the north berths has cost the fleet 'somewhere around \u00a3300,000' in lost landings.", "subjects": ["the harbourmaster", "the fleet"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-correction-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "The closure of the north berths has caused fishing boats to be unable to use the north berths since January.", "subjects": ["north berths"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-correction-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "A public meeting will be held at the harbour office on 2 October.", "subjects": ["the public meeting", "the harbour office"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-correction-001", "run": 2, "outcome": "spurious", "stated": 0.95, "content": "The contract value for the repair is \u00a31.7 million.", "subjects": ["the repair"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-correction-001", "run": 2, "outcome": "matched", "stated": 0.95, "content": "An earlier version of this article said that the repair would cost \u00a32.1 million.", "subjects": ["the article"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "Tamberly Freight runs 260 refrigerated trucks out of three depots in the English Midlands.", "subjects": ["Tamberly Freight"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "Until July 2026 our route planner ran once a night, at 02:00, and drivers received the next day's routes on a printed manifest.", "subjects": ["Tamberly Freight"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "In July we replaced the nightly batch with an event-driven planner that re-plans a route whenever an order is added, cancelled or delayed.", "subjects": ["Tamberly Freight"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-001", "run": 3, "outcome": "spurious", "stated": 0.95, "content": "Each change is published to a message queue, and the planner recomputes only the routes that touch the affected depot.", "subjects": ["Tamberly Freight"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-001", "run": 3, "outcome": "matched", "stated": 0.92, "content": "The results after eight weeks: the average time from an order being placed to it appearing on a driver's route fell from 19 hours to 40 minutes.", "subjects": ["Tamberly Freight"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-001", "run": 3, "outcome": "matched", "stated": 0.92, "content": "Empty running, the share of kilometres driven with no load, dropped from 23 percent to 17 percent.", "subjects": ["Tamberly Freight"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "We considered buying a commercial routing service instead.", "subjects": ["Tamberly Freight"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "We rejected it because none of the three vendors we spoke to supported temperature-zone constraints, and every one of our loads has one.", "subjects": ["Tamberly Freight"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "The migration took a team of five engineers eleven weeks.", "subjects": ["Tamberly Freight"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-001", "run": 3, "outcome": "matched", "stated": 0.94, "content": "Printed manifests are still produced as a backup and will be retired in January 2027.", "subjects": ["Tamberly Freight"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 3, "outcome": "matched", "stated": 0.95, "content": "The Ashcombe Vale public library reopened on Saturday in the former bus depot on Quarry Road.", "subjects": ["Ashcombe Vale public library", "former bus depot on Quarry Road"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 3, "outcome": "matched", "stated": 0.95, "content": "The Ashcombe Vale public library closed on the High Street two years ago because of subsidence.", "subjects": ["Ashcombe Vale public library", "High Street"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 3, "outcome": "matched", "stated": 0.95, "content": "The conversion of the former bus depot into a library cost \u00a33.4 million.", "subjects": ["conversion of the former bus depot into a library"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 3, "outcome": "matched", "stated": 0.95, "content": "\u00a31.1 million of the conversion cost came from a national lottery heritage grant.", "subjects": ["\u00a31.1 million of the conversion cost"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 3, "outcome": "matched", "stated": 0.95, "content": "The remaining \u00a32.3 million of the conversion cost came from the district council.", "subjects": ["remaining \u00a32.3 million of the conversion cost"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 3, "outcome": "matched", "stated": 0.95, "content": "The new library has 1,850 square metres of floor space.", "subjects": ["new library"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 3, "outcome": "matched", "stated": 0.9, "content": "The new library has about twice the floor space of the old building.", "subjects": ["new library", "old building"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 3, "outcome": "matched", "stated": 0.95, "content": "The new library holds 46,000 books.", "subjects": ["new library"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 3, "outcome": "matched", "stated": 0.95, "content": "The new library is open seven days a week, from 9:00 to 20:00 on weekdays and 10:00 to 16:00 at weekends.", "subjects": ["new library"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 3, "outcome": "matched", "stated": 0.95, "content": "The old library closed on Sundays.", "subjects": ["old library"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 3, "outcome": "matched", "stated": 0.95, "content": "The former depot's inspection pit has been kept under a glass floor in the children's section.", "subjects": ["former depot's inspection pit", "children's section"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 3, "outcome": "matched", "stated": 0.85, "content": "Head librarian Ines Calloway said that the library wanted people to see what the building used to be.", "subjects": ["Ines Calloway", "library", "building"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 3, "outcome": "matched", "stated": 0.8, "content": "Around 1,200 people visited the library on the opening day.", "subjects": ["library"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-article-002", "run": 3, "outcome": "matched", "stated": 0.7, "content": "A caf\u00e9 run by the Vale Community Kitchen charity will open in the building in November.", "subjects": ["caf\u00e9 run by the Vale Community Kitchen charity", "building"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-essay-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "Hearthwell Bakery was founded in 1994 as a cooperative by six bakers who had all been laid off when the Morrow & Pike flour mill in Lindfield closed.", "subjects": ["Hearthwell Bakery", "Morrow & Pike flour mill", "Lindfield"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-essay-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "Each member put in \u00a32,000, and the cooperative still pays every member the same hourly wage.", "subjects": ["Hearthwell Bakery"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-essay-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "Its rivals grew. Crumbworth, which opened on the same street in 1996, now supplies 140 supermarkets from a factory outside the town.", "subjects": ["Crumbworth", "Hearthwell Bakery"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-essay-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "Hearthwell has one shop and one oven, a wood-fired oven built in 1994 that is still in daily use.", "subjects": ["Hearthwell Bakery"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-essay-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "The cooperative has twice voted against expansion: in 2008, when a supermarket chain offered a supply contract, and in 2019, when the council offered it a second unit at a reduced rent. Both votes were unanimous.", "subjects": ["Hearthwell Bakery"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-essay-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "Membership has grown only from six to eleven in thirty-two years.", "subjects": ["Hearthwell Bakery"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-essay-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "The shop sells around 900 loaves on a Saturday and closes when they are gone.", "subjects": ["Hearthwell Bakery"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-essay-001", "run": 3, "outcome": "spurious", "stated": 0.6, "content": "I think Hearthwell's refusal to grow is the reason it has survived; a bakery that cannot be bought cheaply is a bakery nobody tries to buy.", "subjects": ["Hearthwell Bakery"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-interview-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "The Corran Point station measures sea level and water temperature at the mouth of the Corran estuary every ten minutes.", "subjects": ["Corran Point station"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-interview-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "The Corran Point station has run without a gap since 1987, which makes it the longest unbroken record on this coast.", "subjects": ["Corran Point station"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-interview-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "Mean sea level at the Corran Point station is 11 centimetres higher than it was in 1987.", "subjects": ["Corran Point station"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-interview-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "Summer surface temperature at the Corran Point station has risen by 1.4 degrees since 1987.", "subjects": ["Corran Point station"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-interview-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "The Corran Estuary Trust owns the Corran Point station.", "subjects": ["Corran Estuary Trust", "Corran Point station"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-interview-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "Kofi Ansah-Brandt is the only full-time staff member of the Corran Point station.", "subjects": ["Kofi Ansah-Brandt", "Corran Point station"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-interview-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "Two volunteers check the sensors at the Corran Point station each week.", "subjects": ["two volunteers", "Corran Point station"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-interview-001", "run": 3, "outcome": "matched", "stated": 0.85, "content": "A salinity sensor will be added to the Corran Point station in October.", "subjects": ["Corran Point station"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-interview-001", "run": 3, "outcome": "matched", "stated": 0.39, "content": "Kofi Ansah-Brandt guesses that the salt wedge is moving further upstream each year at the Corran Point station.", "subjects": ["Kofi Ansah-Brandt", "salt wedge", "Corran Point station"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-interview-001", "run": 3, "outcome": "spurious", "stated": 0.95, "content": "The Corran Point station has no data to show that the salt wedge is moving further upstream each year.", "subjects": ["Corran Point station"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "Skerrin Island Ferries carried 412,000 passengers in the six months to 30 June 2026.", "subjects": ["Skerrin Island Ferries"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "Skerrin Island Ferries carried 371,000 passengers in the six months to 30 June 2025.", "subjects": ["Skerrin Island Ferries"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "Vehicle crossings rose from 58,000 to 63,500.", "subjects": ["Skerrin Island Ferries"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "Revenue for the half was \u00a39.8 million.", "subjects": ["Skerrin Island Ferries"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "Operating profit for the half was \u00a31.2 million.", "subjects": ["Skerrin Island Ferries"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "Fuel was the largest single cost at \u00a32.6 million.", "subjects": ["Skerrin Island Ferries"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "The new hybrid vessel MV Aird Mhor entered service on 3 March.", "subjects": ["MV Aird Mhor"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "MV Aird Mhor now makes eight of the eighteen daily crossings.", "subjects": ["MV Aird Mhor"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "On those crossings, MV Aird Mhor uses 31 percent less diesel than the vessel it replaced.", "subjects": ["MV Aird Mhor"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "Punctuality, measured as departures within five minutes of the timetable, was 94.2 percent.", "subjects": ["Skerrin Island Ferries"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "Forty-one sailings were cancelled for weather in the first half of 2026.", "subjects": ["Skerrin Island Ferries"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "Sixty-six sailings were cancelled for weather in the first half of 2025.", "subjects": ["Skerrin Island Ferries"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-report-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "The board has approved an order for a second hybrid vessel, due in 2028.", "subjects": ["Skerrin Island Ferries"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-incident-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "Card payments through Quillon failed for 47 minutes on 4 September, from 14:12 to 14:59 UTC.", "subjects": ["Quillon"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-incident-001", "run": 3, "outcome": "matched", "stated": 0.9, "content": "That window is confirmed from our gateway logs.", "subjects": ["Quillon"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-incident-001", "run": 3, "outcome": "matched", "stated": 0.75, "content": "Roughly 18,000 transactions were declined; the exact figure will change as merchants report retries.", "subjects": ["Quillon"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-incident-001", "run": 3, "outcome": "matched", "stated": 0.7, "content": "Our leading hypothesis is that a certificate rotation on the card-network link at 14:10 left one of three gateway nodes presenting an expired certificate.", "subjects": ["Quillon"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-incident-001", "run": 3, "outcome": "spurious", "stated": 0.65, "content": "We think this is likely but have not reproduced it.", "subjects": ["Quillon"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-incident-001", "run": 3, "outcome": "spurious", "stated": 0.9, "content": "We have ruled out a database fault: the payments database showed normal latency throughout.", "subjects": ["Quillon"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-incident-001", "run": 3, "outcome": "matched", "stated": 0.7, "content": "It is unclear why traffic was not drained from the affected node automatically.", "subjects": ["Quillon"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-incident-001", "run": 3, "outcome": "matched", "stated": 0.65, "content": "The health check may have been testing the wrong endpoint, but we have not confirmed that.", "subjects": ["Quillon"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-incident-001", "run": 3, "outcome": "spurious", "stated": 0.9, "content": "A full review will be published by 30 September.", "subjects": ["Quillon"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-incident-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "No customer card data was exposed.", "subjects": ["Quillon"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-trial-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "Nine member farms undersowed winter wheat with white clover in 2025 and 2026, leaving a matched strip on each farm without clover.", "subjects": ["Nine member farms"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-trial-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "A tenth farm dropped out after its strip flooded in February and is excluded from every figure here.", "subjects": ["A tenth farm"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-trial-001", "run": 3, "outcome": "matched", "stated": 0.9, "content": "Across the nine farms, nitrogen fertiliser use on the clover strips was 28 percent lower.", "subjects": ["Nitrogen fertiliser use on the clover strips across the nine farms"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-trial-001", "run": 3, "outcome": "matched", "stated": 0.75, "content": "Yields were between 4 percent lower and 2 percent higher than the matched strips.", "subjects": ["Yields on the clover strips compared to the matched strips"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-trial-001", "run": 3, "outcome": "matched", "stated": 0.75, "content": "Earthworm counts were higher on the clover strips on seven of the nine farms.", "subjects": ["Earthworm counts on the clover strips on seven of the nine farms"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-trial-001", "run": 3, "outcome": "matched", "stated": 0.5, "content": "Our best estimate is that a farm adopting undersowing would break even within three seasons.", "subjects": ["A farm adopting undersowing"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-trial-001", "run": 3, "outcome": "matched", "stated": 0.75, "content": "A third year is planned if funding from the regional soils fund is renewed in December.", "subjects": ["A third year of the trial"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "MS 1172 is a two-page letter describing the collapse of the Wheal Tressa mine engine house.", "subjects": ["MS 1172"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "MS 1172 is unsigned.", "subjects": ["MS 1172"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "MS 1172 was given to the Record Office in 1964 by a private donor.", "subjects": ["MS 1172"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "Nothing is known of where MS 1172 was before 1964.", "subjects": ["MS 1172"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 3, "outcome": "matched", "stated": 0.85, "content": "The letter has traditionally been attributed to the mine captain Josiah Penhallow.", "subjects": ["MS 1172"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "The handwriting is close to that of Josiah Penhallow's surviving account books.", "subjects": ["MS 1172"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "The writer refers to 'my men'.", "subjects": ["MS 1172"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 3, "outcome": "matched", "stated": 0.85, "content": "Dr Lowenna Tresize has argued that the letter was written by Josiah Penhallow's daughter, Keziah.", "subjects": ["Dr Lowenna Tresize"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 3, "outcome": "spurious", "stated": 0.85, "content": "Dr Lowenna Tresize has pointed out that the letter uses spellings found in Keziah's letters and never in Josiah's.", "subjects": ["Dr Lowenna Tresize"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 3, "outcome": "spurious", "stated": 0.95, "content": "The Record Office does not take a view.", "subjects": ["Dunmere County Record Office"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "The letter is catalogued as 'attributed to Josiah Penhallow, attribution disputed'.", "subjects": ["MS 1172"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "The paper's watermark dates it to between 1838 and 1846.", "subjects": ["MS 1172"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 3, "outcome": "spurious", "stated": 0.85, "content": "The collapse it describes is usually dated to 1841.", "subjects": ["MS 1172"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-archive-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "No newspaper report of the collapse has been found.", "subjects": ["the collapse of the Wheal Tressa mine engine house"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-forecast-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "The council's spending this year, 2026\u201327, is \u00a314.2 million.", "subjects": ["Ferrisford Town Council"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-forecast-001", "run": 3, "outcome": "matched", "stated": 0.75, "content": "For 2027\u201328 we forecast a gap of between \u00a30.9 million and \u00a31.6 million between expected spending and expected income.", "subjects": ["Ferrisford Town Council"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-forecast-001", "run": 3, "outcome": "spurious", "stated": 0.85, "content": "The width of that range is mostly the leisure centre.", "subjects": ["Ferrisford Town Council", "Leisure centre"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-forecast-001", "run": 3, "outcome": "matched", "stated": 0.7, "content": "If councillors vote in November to close the Ferrisford pool, the gap would fall by about \u00a30.5 million a year.", "subjects": ["Ferrisford Town Council", "Ferrisford pool"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-forecast-001", "run": 3, "outcome": "matched", "stated": 0.75, "content": "If they keep it open, the pool's boiler is expected to need replacing within two years, at an estimated cost of \u00a30.8 million.", "subjects": ["Ferrisford Town Council", "Ferrisford pool"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-forecast-001", "run": 3, "outcome": "matched", "stated": 0.9, "content": "We have assumed council tax rises by the maximum 4.99 percent allowed without a local referendum.", "subjects": ["Ferrisford Town Council"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-forecast-001", "run": 3, "outcome": "spurious", "stated": 0.7, "content": "Each percentage point below that adds roughly \u00a385,000 to the gap.", "subjects": ["Ferrisford Town Council"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-forecast-001", "run": 3, "outcome": "matched", "stated": 0.75, "content": "Our income forecast assumes car-parking revenue returns to its 2019 level.", "subjects": ["Ferrisford Town Council"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-forecast-001", "run": 3, "outcome": "spurious", "stated": 0.6, "content": "Honestly, we think that is optimistic; it has recovered to about 80 percent of that level so far.", "subjects": ["Ferrisford Town Council"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-forecast-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "A draft budget will go to full council on 9 February 2027.", "subjects": ["Ferrisford Town Council"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-correction-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "The repair of Calderhaven's north harbour wall, damaged in the storm of January 2026, will not start until spring 2027.", "subjects": ["Calderhaven's north harbour wall"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-correction-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "Work on the repair of Calderhaven's north harbour wall had been due to start this month (September 2026).", "subjects": ["Calderhaven's north harbour wall"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-correction-001", "run": 3, "outcome": "matched", "stated": 0.8, "content": "The harbour authority blamed the delay on a shortage of granite needed to match the original wall.", "subjects": ["Calderhaven's north harbour wall", "harbour authority"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-correction-001", "run": 3, "outcome": "matched", "stated": 0.8, "content": "The contractor, Pendry Marine, said instead that the authority had not yet approved the final design.", "subjects": ["Pendry Marine", "harbour authority"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-correction-001", "run": 3, "outcome": "spurious", "stated": 0.8, "content": "Pendry Marine said granite was available for the repair of Calderhaven's north harbour wall.", "subjects": ["Pendry Marine", "Calderhaven's north harbour wall"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-correction-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "Fishing boats have been unable to use the north berths since January 2026.", "subjects": ["Fishing boats", "north berths"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-correction-001", "run": 3, "outcome": "matched", "stated": 0.8, "content": "The harbourmaster estimates the closure of the north berths has cost the fleet 'somewhere around \u00a3300,000' in lost landings.", "subjects": ["harbourmaster", "north berths", "fleet"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-correction-001", "run": 3, "outcome": "matched", "stated": 0.75, "content": "The harbourmaster said the figure of \u00a3300,000 is a rough one.", "subjects": ["harbourmaster"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-correction-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "A public meeting will be held at the harbour office on 2 October.", "subjects": ["public meeting", "harbour office"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-correction-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "The contract value for the repair of Calderhaven's north harbour wall is \u00a31.7 million.", "subjects": ["Calderhaven's north harbour wall"]}
{"provider_model": "local:qwen3:14b", "extractor_version": "0.16.0", "suite": "prose-holdout-001", "suite_version": "0.1.0", "case": "hold-hedged-correction-001", "run": 3, "outcome": "matched", "stated": 0.95, "content": "An earlier version of this article said the repair would cost \u00a32.1 million.", "subjects": ["article", "repair of Calderhaven's north harbour wall"]}
