File size: 2,675 Bytes
344a8e9 0410848 344a8e9 0410848 | 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 64 65 66 67 68 69 70 71 72 73 74 75 76 | {
"name": "ScienceQA",
"release_date": "2022-09-20",
"subsets": {
"main": {
"language": [
"en"
],
"modalities": [
"single_image_start",
"text"
],
"task_type": "multiple_choice_qa",
"score_pipeline": [
"exact-match",
"rule-match"
],
"score_protocol": {
"reference": "lmms-eval@lmms_eval/tasks/scienceqa/utils.py:34-44 sqa_process_results — llava-style regex parse of the option letter, exact_match metric, no LLM. Matches official lupantech/ScienceQA get_pred_idx rule-based scoring. Confirmed in clone (pred compared to option letter, ^[A-Z]\\. pattern).",
"note": "Mixed-modality (text-only rows have no image); all rows are MCQ, graded by exact letter match."
},
"prompt_template": "{% if has_image %}<image>{% endif %}{% if hint %}Context: {{ hint }}\n{% endif %}Question: {{ question }}\nOptions:\n{% for k, v in options.items() %}({{ k }}) {{ v }}\n{% endfor %}Answer with the option letter from the given choices.",
"mapping_from_source": {
"media": {
"from": "image",
"type": "list",
"min_items": 0,
"max_items": 1
},
"id": {
"from": "id"
},
"question": {
"from": "question"
},
"answer": {
"from": "answer",
"optional": true
},
"hint": {
"from": "hint",
"optional": true
},
"options": {
"from": "choices",
"optional": true,
"note": "list source values are normalized to {A,B,...} dict"
},
"extra": {
"has_image": {
"from": "has_image"
},
"subject": {
"from": "subject"
},
"grade": {
"from": "grade"
}
},
"source": {
"format": "json",
"url": {
"train": "https://huggingface.co/datasets/derek-thomas/ScienceQA",
"validation": "https://huggingface.co/datasets/derek-thomas/ScienceQA",
"test": "https://huggingface.co/datasets/derek-thomas/ScienceQA"
}
}
},
"prompt_template_source": {
"origin": "official",
"reference": "https://github.com/EvolvingLMMs-Lab/lmms-eval/blob/main/lmms_eval/tasks/scienceqa/scienceqa.yaml (canonical post_prompt='\\nAnswer with the option's letter from the given choices directly.')",
"notes": "Tier 4: lmms-eval ScienceQA canonical evaluation prompt; supports has_image/hint conditional rendering for the mixed-modality benchmark."
}
}
}
} |