File size: 2,564 Bytes
e8ac1ae 0734445 e8ac1ae 0734445 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 | {
"name": "3DSRBench",
"release_date": "2024-12-01",
"subsets": {
"main": {
"language": [
"en"
],
"modalities": [
"single_image_start"
],
"task_type": "multiple_choice_qa",
"score_pipeline": [
"rule-match",
"llm-match"
],
"score_protocol": {
"reference": "VLMEvalKit@vlmeval/dataset/image_mcq.py:2662-2740 — _3DSRBench.evaluate uses mcq_vanilla_eval (exact/heuristic option inference first, GPT extraction fallback when a judge model is configured), then flip and circular aggregations; the authors distribute the VLMEvalKit-format circular TSV as the official eval path (huggingface.co/datasets/ccvl/3DSRBench).",
"note": "Official headline is FlipEval accuracy: hit for a question is the product over the original and '-flip' rows (image_mcq.py:2707-2716), plus circular-consistency accuracy (2717-2726); a per-sample mean reproduces only the 'vanilla' number. The mm-eval copy ships the flip rows (ids '<id>-flip'), so pair aggregation remains possible downstream. lmms-eval 3dsrbench (tasks/3dsrbench/utils.py) grades with pure regex extraction, no LLM."
},
"prompt_template": "<image>{% if hint %}Hint: {{ hint }}\n{% endif %}Question: {{ question }}\nOptions:\n{% for k, v in options.items() %}{{ k }}. {{ v }}\n{% endfor %}Please select the correct answer from the options above. ",
"mapping_from_source": {
"media": {
"from": "image",
"type": "list",
"min_items": 1,
"max_items": 1
},
"id": {
"from": "id"
},
"question": {
"from": "question"
},
"answer": {
"from": "answer",
"optional": true
},
"options": {
"from": "options",
"optional": true,
"note": "list source values are normalized to {A,B,...} dict"
},
"extra": {
"category": {
"from": "category"
}
},
"source": {
"format": "json",
"url": {
"test": "https://huggingface.co/datasets/ccvl/3DSRBench"
}
}
},
"prompt_template_source": {
"origin": "official",
"reference": "https://github.com/EvolvingLMMs-Lab/lmms-eval/blob/main/lmms_eval/tasks/3dsrbench/utils.py (3dsrbench_doc_to_text — canonical MCQ Hint:/Question:/Options:/Please select... trailer)",
"notes": "Tier 4: lmms-eval 3DSRBench canonical evaluation prompt."
}
}
}
} |