{"_id": {"$oid": "6a26c67c3572c06bc1455cf3"}, "test_id": "T0809", "benchmark": "magazine_pages", "date": "2026-03-23", "tags": {"century": [20], "document-type": ["newspaper-page"], "language": ["en"], "layout": ["prose", "columns"], "script": ["latin"], "task": ["document-understanding"], "writing": ["printed"]}, "contributors": ["lea_kasper", "sorin_marti", "arno_bosse"], "hidden": false, "config": {"provider": "mistral", "model": "ministral-8b-2512", "dataclass": "MagazinePage", "temperature": 0.0, "role_description": "You are a historian with expertise in analyzing and summarizing historical magazine pages", "prompt_file": "prompt.txt", "legacy_test": false}, "prompt": "Extract all advertisements and return their bounding boxes.\nThe original size of the page is {width} x {height} pixels.", "results": {"text": "{\"advertisements\":[{\"box\":[100.0,50.0,450.0,800.0]},{\"box\":[1500.0,100.0,2000.0,1200.0]},{\"box\":[200.0,2000.0,600.0,3200.0]},{\"box\":[1200.0,2500.0,1800.0,3400.0]},{\"box\":[1800.0,100.0,2300.0,800.0]},{\"box\":[100.0,1800.0,400.0,2500.0]},{\"box\":[1500.0,2800.0,2000.0,3400.0]},{\"box\":[2200.0,100.0,2470.0,400.0]}]}", "model": "ministral-8b-2512", "provider": "mistral", "finish_reason": "stop", "usage": {"input_tokens": 254, "output_tokens": 475, "total_tokens": 729, "input_cost_usd": 3.81e-05, "output_cost_usd": 7.125e-05, "estimated_cost_usd": 0.00010935}, "duration": 3.7965285778045654, "timestamp": "2026-03-23T22:28:58.818376", "parsed": {"$defs": {"Advertisement": {"properties": {"box": {"description": "Bounding box of the advertisement as [x0, y0, x1, y1] pixel coordinates, where (x0, y0) is the top-left corner and (x1, y1) is the bottom-right corner.", "items": {"type": "number"}, "title": "Box", "type": "array"}}, "required": ["box"], "title": "Advertisement", "type": "object"}}, "properties": {"advertisements": {"items": {"$ref": "#/$defs/Advertisement"}, "title": "Advertisements"}}, "required": ["advertisements"], "title": "MagazinePage", "type": "object", "advertisements": [{"box": [100, 50, 450, 800]}, {"box": [1500, 100, 2000, 1200]}, {"box": [200, 2000, 600, 3200]}, {"box": [1200, 2500, 1800, 3400]}, {"box": [1800, 100, 2300, 800]}, {"box": [100, 1800, 400, 2500]}, {"box": [1500, 2800, 2000, 3400]}, {"box": [2200, 100, 2470, 400]}]}, "conversation_id": "f4dc3d81-7cff-4d9a-9dec-f73dffe1accb", "score": {"f1": 0.0, "precision": 0.0, "recall": 0.0, "mean_iou": 0.0, "true_positives": 0, "false_positives": 8, "false_negatives": 1, "iou_threshold": 0.5}, "raw_response": "{\"id\":\"350f92343eeb4285981e3ff8660e4b84\",\"object\":\"chat.completion\",\"model\":\"ministral-8b-2512\",\"usage\":{\"prompt_tokens\":254,\"completion_tokens\":475,\"total_tokens\":729,\"prompt_tokens_details\":{\"cached_tokens\":228}},\"created\":1774301335,\"choices\":[{\"index\":0,\"message\":{\"role\":\"assistant\",\"content\":\"{\\n  \\\"$defs\\\": {\\n    \\\"Advertisement\\\": {\\n      \\\"properties\\\": {\\n        \\\"box\\\": {\\n          \\\"description\\\": \\\"Bounding box of the advertisement as [x0, y0, x1, y1] pixel coordinates, where (x0, y0) is the top-left corner and (x1, y1) is the bottom-right corner.\\\",\\n          \\\"items\\\": {\\n            \\\"type\\\": \\\"number\\\"\\n          },\\n          \\\"title\\\": \\\"Box\\\",\\n          \\\"type\\\": \\\"array\\\"\\n        }\\n      },\\n      \\\"required\\\": [\\n        \\\"box\\\"\\n      ],\\n      \\\"title\\\": \\\"Advertisement\\\",\\n      \\\"type\\\": \\\"object\\\"\\n    }\\n  },\\n  \\\"properties\\\": {\\n    \\\"advertisements\\\": {\\n      \\\"items\\\": {\\n        \\\"$ref\\\": \\\"#/$defs/Advertisement\\\"\\n      },\\n      \\\"title\\\": \\\"Advertisements\\\"\\n    }\\n  },\\n  \\\"required\\\": [\\n    \\\"advertisements\\\"\\n  ],\\n  \\\"title\\\": \\\"MagazinePage\\\",\\n  \\\"type\\\": \\\"object\\\",\\n  \\\"advertisements\\\": [\\n    {\\n      \\\"box\\\": [100, 50, 450, 800]\\n    },\\n    {\\n      \\\"box\\\": [1500, 100, 2000, 1200]\\n    },\\n    {\\n      \\\"box\\\": [200, 2000, 600, 3200]\\n    },\\n    {\\n      \\\"box\\\": [1200, 2500, 1800, 3400]\\n    },\\n    {\\n      \\\"box\\\": [1800, 100, 2300, 800]\\n    },\\n    {\\n      \\\"box\\\": [100, 1800, 400, 2500]\\n    },\\n    {\\n      \\\"box\\\": [1500, 2800, 2000, 3400]\\n    },\\n    {\\n      \\\"box\\\": [2200, 100, 2470, 400]\\n    }\\n  ]\\n}\",\"tool_calls\":null,\"prefix\":false},\"finish_reason\":\"stop\"}]}"}, "scoring": {"f1": 0.003, "precision": 0.003, "recall": 0.003, "mean_iou": 0.015, "true_positives": 1, "false_positives": 373, "false_negatives": 125, "num_pages": 46, "cost_summary": {"total_input_tokens": 11684, "total_output_tokens": 24062, "total_tokens": 35746, "input_cost_usd": 0.0017525999999999978, "output_cost_usd": 0.0036092999999999997, "total_cost_usd": 0.0053618999999999984}}, "normalized_score": 0.3}