複数選択肢質問応答(MCQA)
複数選択質問応答(MCQA)は、自然言語処理のサブタスクの一つで、モデルが提供された候補オプションとサポート情報に基づいて与えられた質問に対する最適な答えを予測することを要求します。このタスクは、モデルの理解力と推論能力を評価することを目指しており、知能教育、オンライン評価、知識検索など、幅広い応用価値を持っています。
MedMCQA
Meditron-70B (CoT + SC)
BIG-bench (Hyperbaton)
BIG-bench (Movie Recommendation)
BIG-bench (Navigate)
BIG-bench (Ruin Names)
MMLU (College Biology)
Chinchilla (few-shot, k=5)
MMLU (Medical Genetics)
MMLU (Professional medicine)
MMLU (Elementary Mathematics)
Chinchilla (few-shot, k=5)
MMLU (High School Biology)
Chinchilla (few-shot, k=5)
MMLU (College Chemistry)
Chinchilla (few-shot, k=5)
MMLU (High School Mathematics)
GAL 120B (zero-shot)
MMLU (Electrical Engineer)
GAL 120B (zero-shot)
MMLU (College Physics)
MMLU (Formal Logic)
Gopher (few-shot, k=5)
MMLU (High School Statistics)
MMLU (Abstract Algebra)
GAL 30B (zero-shot)
MMLU (Econometrics)
Gopher (few-shot, k=5)
MMLU (High School Computer Science)
GAL 120B (zero-shot)
MMLU (College Mathematics)
GAL 120B (zero-shot)
MMLU (Astronomy)
Chinchilla (few-shot, k=5)
MMLU (High School Chemistry)
Chinchilla (few-shot, k=5)
MMLU (College Computer Science)
Chinchilla (few-shot, k=5)
MMLU (High School Physics)
BIG-bench (Novel Concepts)
MMLU (Machine Learning)
Chinchilla (few-shot, k=5)
IndicGLUE WSTP Pa
MMLU (Clinical Knowledge)
MMLU (Anatomy)
Med-PaLM 2 (ER)
MMLU (College Medicine)
FrenchMedMCQA
CamemBERT