OVEN GRPO sweep (Qwen3-VL-4B)
Collection
GRPO LoRA fine-tunes of Qwen3-VL-4B on OVEN (verl), from my MSc thesis. Mostly negative results across reward, prompt, and data axes. • 14 items • Updated
data_source stringclasses 1
value | prompt listlengths 2 2 | ability stringclasses 1
value | reward_model dict | images listlengths 1 1 | extra_info dict |
|---|---|---|---|---|---|
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Miniature Pinscher"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_01632284.JPEG"
}
] | {
"index": 0,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_01632859",
"image_id": "oven_01632284",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_01632284.jpg",
"entity_id": "Q38990",
"answer": "Miniature Pinscher",
"entity_text": "Miniature Pinscher",
"questi... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Air gun"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_02055626.JPEG"
}
] | {
"index": 1,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_02056201",
"image_id": "oven_02055626",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_02055626.jpg",
"entity_id": "Q193427",
"answer": "Air gun",
"entity_text": "Air gun",
"question": "what is this we... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Alvastra Abbey"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_00110632.jpg"
}
] | {
"index": 2,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_00111207",
"image_id": "oven_00110632",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_00110632.jpg",
"entity_id": "Q1756189",
"answer": "Alvastra Abbey",
"entity_text": "Alvastra Abbey",
"question": "... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Church of the Intercession on the Nerl"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_00029513.jpg"
}
] | {
"index": 3,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_00030088",
"image_id": "oven_00029513",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_00029513.jpg",
"entity_id": "Q1051458",
"answer": "Church of the Intercession on the Nerl",
"entity_text": "Church o... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Ring-necked snake"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_00649213.JPEG"
}
] | {
"index": 4,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_00649788",
"image_id": "oven_00649213",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_00649213.jpg",
"entity_id": "Q2704229",
"answer": "Ring-necked snake",
"entity_text": "Ring-necked snake",
"questi... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Reflex camera"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_03432066.JPEG"
}
] | {
"index": 5,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_03432641",
"image_id": "oven_03432066",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_03432066.jpg",
"entity_id": "Q2858617",
"answer": "Reflex camera",
"entity_text": "Reflex camera",
"question": "wh... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Newsagent's shop"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_03197794.JPEG"
}
] | {
"index": 6,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_03198369",
"image_id": "oven_03197794",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_03197794.jpg",
"entity_id": "Q1528905",
"answer": "Newsagent's shop",
"entity_text": "Newsagent's shop",
"question... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Penguin"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_01483581.JPEG"
}
] | {
"index": 7,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_01484156",
"image_id": "oven_01483581",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_01483581.jpg",
"entity_id": "Q9147",
"answer": "Penguin",
"entity_text": "Penguin",
"question": "what is the categ... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Stem rust"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_04873254.JPEG"
}
] | {
"index": 8,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_04873829",
"image_id": "oven_04873254",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_04873254.jpg",
"entity_id": "Q1342327",
"answer": "Stem rust",
"entity_text": "Stem rust",
"question": "which cate... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Carioca Aqueduct"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_00010450.jpg"
}
] | {
"index": 9,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_00011025",
"image_id": "oven_00010450",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_00010450.jpg",
"entity_id": "Q1579892",
"answer": "Carioca Aqueduct",
"entity_text": "Carioca Aqueduct",
"question... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Bluegill"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_00436525.JPEG"
}
] | {
"index": 10,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_00437100",
"image_id": "oven_00436525",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_00436525.jpg",
"entity_id": "Q1148148",
"answer": "Bluegill",
"entity_text": "Bluegill",
"question": "what is thi... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Eucalyptus tereticornis"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_04678869.JPEG"
}
] | {
"index": 11,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_04679444",
"image_id": "oven_04678869",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_04678869.jpg",
"entity_id": "Q3006305",
"answer": "Eucalyptus tereticornis",
"entity_text": "Eucalyptus tereticorni... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Potala Palace"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_00117487.jpg"
}
] | {
"index": 12,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_00118062",
"image_id": "oven_00117487",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_00117487.jpg",
"entity_id": "Q71229",
"answer": "Potala Palace",
"entity_text": "Potala Palace",
"question": "wha... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Scavenger"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_01093813.JPEG"
}
] | {
"index": 13,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_01094388",
"image_id": "oven_01093813",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_01093813.jpg",
"entity_id": "Q172867",
"answer": "Scavenger",
"entity_text": "Scavenger",
"question": "what is th... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Electrical network"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_02460802.JPEG"
}
] | {
"index": 14,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_02461377",
"image_id": "oven_02460802",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_02460802.jpg",
"entity_id": "Q132629",
"answer": "Electrical network",
"entity_text": "Electrical network",
"ques... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Wat Saket"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_00011991.jpg"
}
] | {
"index": 15,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_00012566",
"image_id": "oven_00011991",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_00011991.jpg",
"entity_id": "Q2063818",
"answer": "Wat Saket",
"entity_text": "Wat Saket",
"question": "what is t... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Toque"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_03308807.JPEG"
}
] | {
"index": 16,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_03309382",
"image_id": "oven_03308807",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_03308807.jpg",
"entity_id": "Q2309028",
"answer": "Toque",
"entity_text": "Toque",
"question": "what is this cap?... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Bookmark"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_02254534.JPEG"
}
] | {
"index": 17,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_02255109",
"image_id": "oven_02254534",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_02254534.jpg",
"entity_id": "Q1065044",
"answer": "Bookmark",
"entity_text": "Bookmark",
"question": "what is thi... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Bean"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_04148109.JPEG"
}
] | {
"index": 18,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_04148684",
"image_id": "oven_04148109",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_04148109.jpg",
"entity_id": "Q379813",
"answer": "Bean",
"entity_text": "Bean",
"question": "what is this common ... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Verbesina encelioides"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_00519937.JPEG"
}
] | {
"index": 19,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_00520512",
"image_id": "oven_00519937",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_00519937.jpg",
"entity_id": "Q7160909",
"answer": "Verbesina encelioides",
"entity_text": "Verbesina encelioides",
... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Moquette"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_03164291.JPEG"
}
] | {
"index": 20,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_03164866",
"image_id": "oven_03164291",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_03164291.jpg",
"entity_id": "Q3055724",
"answer": "Moquette",
"entity_text": "Moquette",
"question": "what is thi... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "English Setter"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_01598504.JPEG"
}
] | {
"index": 21,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_01599079",
"image_id": "oven_01598504",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_01598504.jpg",
"entity_id": "Q38309",
"answer": "English Setter",
"entity_text": "English Setter",
"question": "w... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Blue Mountain (New York)"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_00058170.jpg"
}
] | {
"index": 22,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_00058745",
"image_id": "oven_00058170",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_00058170.jpg",
"entity_id": "Q4929493",
"answer": "Blue Mountain (New York)",
"entity_text": "Blue Mountain (New Yo... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Gatehouse"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_02830287.JPEG"
}
] | {
"index": 23,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_02830862",
"image_id": "oven_02830287",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_02830287.jpg",
"entity_id": "Q277760",
"answer": "Gatehouse",
"entity_text": "Gatehouse",
"question": "what is th... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Parlour"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_00225260.JPEG"
}
] | {
"index": 24,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_00225835",
"image_id": "oven_00225260",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_00225260.jpg",
"entity_id": "Q10550521",
"answer": "Parlour",
"entity_text": "Parlour",
"question": "what is this... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Heverlee War Cemetery"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_00054471.jpg"
}
] | {
"index": 25,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_00055046",
"image_id": "oven_00054471",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_00054471.jpg",
"entity_id": "Q5748310",
"answer": "Heverlee War Cemetery",
"entity_text": "Heverlee War Cemetery",
... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Western dress codes"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_02638247.JPEG"
}
] | {
"index": 26,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_02638822",
"image_id": "oven_02638247",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_02638247.jpg",
"entity_id": "Q28978167",
"answer": "Western dress codes",
"entity_text": "Western dress codes",
"... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Khan Mohammad Mridha Mosque"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_00027067.jpg"
}
] | {
"index": 27,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_00027642",
"image_id": "oven_00027067",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_00027067.jpg",
"entity_id": "Q2667438",
"answer": "Khan Mohammad Mridha Mosque",
"entity_text": "Khan Mohammad Mrid... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "O-ring"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_03226445.JPEG"
}
] | {
"index": 28,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_03227020",
"image_id": "oven_03226445",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_03226445.jpg",
"entity_id": "Q942417",
"answer": "O-ring",
"entity_text": "O-ring",
"question": "what is this typ... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Chintz"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_02445348.JPEG"
}
] | {
"index": 29,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_02445923",
"image_id": "oven_02445348",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_02445348.jpg",
"entity_id": "Q1074462",
"answer": "Chintz",
"entity_text": "Chintz",
"question": "what is this pr... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Momordica balsamina"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_04632726.JPEG"
}
] | {
"index": 30,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_04633301",
"image_id": "oven_04632726",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_04632726.jpg",
"entity_id": "Q2672305",
"answer": "Momordica balsamina",
"entity_text": "Momordica balsamina",
"q... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Sansevieria"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_04715960.JPEG"
}
] | {
"index": 31,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_04716535",
"image_id": "oven_04715960",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_04715960.jpg",
"entity_id": "Q312160",
"answer": "Sansevieria",
"entity_text": "Sansevieria",
"question": "which ... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Knee pad"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_03016883.JPEG"
}
] | {
"index": 32,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_03017458",
"image_id": "oven_03016883",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_03016883.jpg",
"entity_id": "Q681515",
"answer": "Knee pad",
"entity_text": "Knee pad",
"question": "what is this... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Riboflavin"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_04920007.JPEG"
}
] | {
"index": 33,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_04920582",
"image_id": "oven_04920007",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_04920007.jpg",
"entity_id": "Q130365",
"answer": "Riboflavin",
"entity_text": "Riboflavin",
"question": "what is ... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Petri dish"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_03296137.JPEG"
}
] | {
"index": 34,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_03296712",
"image_id": "oven_03296137",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_03296137.jpg",
"entity_id": "Q81714",
"answer": "Petri dish",
"entity_text": "Petri dish",
"question": "what is t... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Lavalier"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_03039018.JPEG"
}
] | {
"index": 35,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_03039593",
"image_id": "oven_03039018",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_03039018.jpg",
"entity_id": "Q4246425",
"answer": "Lavalier",
"entity_text": "Lavalier",
"question": "what is thi... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Alexanderplatz"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_00095708.jpg"
}
] | {
"index": 36,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_00096283",
"image_id": "oven_00095708",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_00095708.jpg",
"entity_id": "Q152783",
"answer": "Alexanderplatz",
"entity_text": "Alexanderplatz",
"question": "... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Auberge de Castille"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_00084456.jpg"
}
] | {
"index": 37,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_00085031",
"image_id": "oven_00084456",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_00084456.jpg",
"entity_id": "Q2499012",
"answer": "Auberge de Castille",
"entity_text": "Auberge de Castille",
"q... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "National Theatre of Panama"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_00113892.jpg"
}
] | {
"index": 38,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_00114467",
"image_id": "oven_00113892",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_00113892.jpg",
"entity_id": "Q9082179",
"answer": "National Theatre of Panama",
"entity_text": "National Theatre of... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Geranium pratense"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_04768614.JPEG"
}
] | {
"index": 39,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_04769189",
"image_id": "oven_04768614",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_04768614.jpg",
"entity_id": "Q157585",
"answer": "Geranium pratense",
"entity_text": "Geranium pratense",
"questi... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Sea lion"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_01504764.JPEG"
}
] | {
"index": 40,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_01505339",
"image_id": "oven_01504764",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_01504764.jpg",
"entity_id": "Q3887135",
"answer": "Sea lion",
"entity_text": "Sea lion",
"question": "which categ... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Dhanmondi Shahi Eidgah"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_00119498.jpg"
}
] | {
"index": 41,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_00120073",
"image_id": "oven_00119498",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_00119498.jpg",
"entity_id": "Q6932152",
"answer": "Dhanmondi Shahi Eidgah",
"entity_text": "Dhanmondi Shahi Eidgah"... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Mucor"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_04838920.JPEG"
}
] | {
"index": 42,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_04839495",
"image_id": "oven_04838920",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_04838920.jpg",
"entity_id": "Q310801",
"answer": "Mucor",
"entity_text": "Mucor",
"question": "what is the catego... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Carpenter bee"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_01743879.JPEG"
}
] | {
"index": 43,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_01744454",
"image_id": "oven_01743879",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_01743879.jpg",
"entity_id": "Q1165550",
"answer": "Carpenter bee",
"entity_text": "Carpenter bee",
"question": "w... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Shift key"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_03549515.JPEG"
}
] | {
"index": 44,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_03550090",
"image_id": "oven_03549515",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_03549515.jpg",
"entity_id": "Q827156",
"answer": "Shift key",
"entity_text": "Shift key",
"question": "what is th... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Black rat"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_00686912.JPEG"
}
] | {
"index": 45,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_00687487",
"image_id": "oven_00686912",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_00686912.jpg",
"entity_id": "Q106133",
"answer": "Black rat",
"entity_text": "Black rat",
"question": "which cate... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "West Voe of Sumburgh"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_00101466.jpg"
}
] | {
"index": 46,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_00102041",
"image_id": "oven_00101466",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_00101466.jpg",
"entity_id": "Q7986963",
"answer": "West Voe of Sumburgh",
"entity_text": "West Voe of Sumburgh",
... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Comfort food"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_03999906.JPEG"
}
] | {
"index": 47,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_04000481",
"image_id": "oven_03999906",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_03999906.jpg",
"entity_id": "Q7887327",
"answer": "Comfort food",
"entity_text": "Comfort food",
"question": "wha... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Jeep Patriot"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_00007220.jpg"
}
] | {
"index": 48,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_00007795",
"image_id": "oven_00007220",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_00007220.jpg",
"entity_id": "Q1686221",
"answer": "Jeep Patriot",
"entity_text": "Jeep Patriot",
"question": "wha... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Phalanger"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_01377814.JPEG"
}
] | {
"index": 49,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_01378389",
"image_id": "oven_01377814",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_01377814.jpg",
"entity_id": "Q1413190",
"answer": "Phalanger",
"entity_text": "Phalanger",
"question": "what is t... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Hôtel de Soubise"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_00055170.jpg"
}
] | {
"index": 50,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_00055745",
"image_id": "oven_00055170",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_00055170.jpg",
"entity_id": "Q917337",
"answer": "Hôtel de Soubise",
"entity_text": "Hôtel de Soubise",
"question... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Papaver heterophyllum"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_00661513.JPEG"
}
] | {
"index": 51,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_00662088",
"image_id": "oven_00661513",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_00661513.jpg",
"entity_id": "Q25407894",
"answer": "Papaver heterophyllum",
"entity_text": "Papaver heterophyllum",... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Tank"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_03720996.JPEG"
}
] | {
"index": 52,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_03721571",
"image_id": "oven_03720996",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_03720996.jpg",
"entity_id": "Q12876",
"answer": "Tank",
"entity_text": "Tank",
"question": "what is this vehicle ... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Pippin"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_04167290.JPEG"
}
] | {
"index": 53,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_04167865",
"image_id": "oven_04167290",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_04167290.jpg",
"entity_id": "Q8448057",
"answer": "Pippin",
"entity_text": "Pippin",
"question": "what is this Wi... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Car door"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_02377559.JPEG"
}
] | {
"index": 54,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_02378134",
"image_id": "oven_02377559",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_02377559.jpg",
"entity_id": "Q1276284",
"answer": "Car door",
"entity_text": "Car door",
"question": "what is thi... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Autun Cathedral"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_00106943.jpg"
}
] | {
"index": 55,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_00107518",
"image_id": "oven_00106943",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_00106943.jpg",
"entity_id": "Q611944",
"answer": "Autun Cathedral",
"entity_text": "Autun Cathedral",
"question":... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "National Centre for the Performing Arts (China)"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_00041603.jpg"
}
] | {
"index": 56,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_00042178",
"image_id": "oven_00041603",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_00041603.jpg",
"entity_id": "Q860188",
"answer": "National Centre for the Performing Arts (China)",
"entity_text": ... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Chlamydosaurus"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_01256127.JPEG"
}
] | {
"index": 57,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_01256702",
"image_id": "oven_01256127",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_01256127.jpg",
"entity_id": "Q501566",
"answer": "Chlamydosaurus",
"entity_text": "Chlamydosaurus",
"question": "... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Arecaceae"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_04735665.JPEG"
}
] | {
"index": 58,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_04736240",
"image_id": "oven_04735665",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_04735665.jpg",
"entity_id": "Q14080",
"answer": "Arecaceae",
"entity_text": "Arecaceae",
"question": "which categ... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Acura TSX"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_00006823.jpg"
}
] | {
"index": 59,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_00007398",
"image_id": "oven_00006823",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_00006823.jpg",
"entity_id": "Q2455196",
"answer": "Acura TSX",
"entity_text": "Acura TSX",
"question": "what is t... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Lovell Telescope"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_00033146.jpg"
}
] | {
"index": 60,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_00033721",
"image_id": "oven_00033146",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_00033146.jpg",
"entity_id": "Q555130",
"answer": "Lovell Telescope",
"entity_text": "Lovell Telescope",
"question... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Antenna (radio)"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_03405414.JPEG"
}
] | {
"index": 61,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_03405989",
"image_id": "oven_03405414",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_03405414.jpg",
"entity_id": "Q131214",
"answer": "Antenna (radio)",
"entity_text": "Antenna (radio)",
"question":... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Powder horn"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_03364651.JPEG"
}
] | {
"index": 62,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_03365226",
"image_id": "oven_03364651",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_03364651.jpg",
"entity_id": "Q7236012",
"answer": "Powder horn",
"entity_text": "Powder horn",
"question": "what ... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Miso soup"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_00199593.jpg"
}
] | {
"index": 63,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_00200168",
"image_id": "oven_00199593",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_00199593.jpg",
"entity_id": "Q471866",
"answer": "Miso soup",
"entity_text": "Miso soup",
"question": "what is th... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Hyundai Grandeur"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_00008079.jpg"
}
] | {
"index": 64,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_00008654",
"image_id": "oven_00008079",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_00008079.jpg",
"entity_id": "Q483860",
"answer": "Hyundai Grandeur",
"entity_text": "Hyundai Grandeur",
"question... |
oven_taxonomy_reasoning | [
{
"role": "system",
"content": "You are a visual entity recognition expert. Your task is to identify the most specific Wikidata entity in an image. For each problem, you will be given an image, a question, and several candidate solutions. Some candidates may be correct, others may contain errors. Evaluate e... | open_world_image_classification | {
"style": "rule",
"ground_truth": "Reticulated python"
} | [
{
"image": "file:///leonardo_work/EUHPC_D33_243/oven/data/images/oven_01279219.JPEG"
}
] | {
"index": 65,
"split": "train",
"data_split": "entity_train",
"data_id": "oven_entity_train_01279794",
"image_id": "oven_01279219",
"image_path": "/leonardo_work/EUHPC_D33_243/oven/oven_01279219.jpg",
"entity_id": "Q216136",
"answer": "Reticulated python",
"entity_text": "Reticulated python",
"ques... |
End of preview. Expand in Data Studio
OVEN GRPO Training Data
Training data behind my MSc thesis GRPO experiments on OVEN open-domain visual entity recognition with Qwen3-VL-4B. This is the companion to the published GRPO adapters: the parquets, taxonomy oracle, and self-aggregation traces they were trained on.
Layout
| path | contents |
|---|---|
taxonomy/chains_labels.jsonl |
Wikidata P279 (subclass-of) chains per OVEN entity: the label hierarchy used by every hierarchical metric |
taxonomy/chains_descs.jsonl |
the same chains with node descriptions (rich-descriptions judge prompt) |
taxonomy/oven_taxonomy_index.json |
prebuilt node index the evaluation + reward code loads at runtime |
grpo/traversal_768/ |
headline run: traversal prompt, shaped reward (aligned, qid-balanced, 2k) |
grpo/aggregation05/, grpo/aggregation08/ |
aggregation-prompt runs (one-shot, 3k) |
grpo/traversal08/ |
traversal run (one-shot, 3k) |
grpo/trace_250k/ |
master 250k self-aggregation-trace training set (+ val, val_unseen) (TIER=full) |
rsa_candidates/ |
N=16 K=4 T=1 recursive-self-aggregation solution traces (+ generation_metadata.json) used to build the aggregation prompts (TIER=full) |
grpo/exploratory/ |
v2/v3/v4 reward/prompt/format ablations (TIER=all) |
Format
The grpo/* parquets are in the verl RL schema,
one row per example: a chat prompt (system + user), an images column (OVEN image),
reward_model = {style: rule, ground_truth: <entity>}, ability: classification, and
extra_info. Each run dir also carries manifest.json and difficulty-stratified subset
parquets (easy / unlockable / inaccessible). Consumed directly by the verl GRPO trainer
with data.image_key=images.
Provenance & licensing
MSc thesis, University of Trento (Juan Camacho Mohedano). Benchmark: OVEN (Hu et al.,
ICCV 2023); the parquet images embed OVEN's Wikipedia images. The taxonomy chains derive
from Wikidata (CC0). The reward and dataset builders live in the
oven-mllm-eval repository.
- Downloads last month
- 31