A 34-item BIG-bench task that picks which English proverb best fits a short anecdote, with three to six choices.
unassessed
| Category | reasoning |
|---|---|
| Subcategory | BIG-bench story-to-proverb multiple choice (34 items, 3–6 options) |
| Page status | unknown |
| Metric | multiple_choice_grade |
| Direction | higher_is_better |
| Unit | % |
| Dataset size | 34 |
| Dataset licence | Apache-2.0 |
| Publisher | Google (BIG-bench collaboration) |
english_proverbs gives a short English story about people and outcomes, then asks which listed proverb best applies. Daniel Moseguí González is the listed author. Each input ends with "Which of the following proverbs best apply to this situation?" Options are conventional English sayings. Six of the 34 examples mark two or three gold proverbs. The skill is matching a narrative to a conventional figurative meaning, not writing a proverb and not [abstract_narrative_understanding](abstract_narrative_understanding.md), which maps crowdsourced stories onto a 100-proverb list with 4, 9, or 99 distractors.
Multiple choice with 3–6 options (28 four-way, four five-way, one three-way, one six-way), preferred metric multiple_choice_grade. No task_prefix in task.json; the question is inside each input. Canary GUID embedded.
No model card in ModelSpec reports this benchmark yet.