jhu-clsp/

AnaloBench

How should we understand each split name? Here is your guide:
This dataset includes two Analogy Tasks:

T1: Classification task: Task 1 is a smaller task that requires models to select the story most analogous from four options. T2: Retrieval task Task 2 is a larger task where models are tasked with choosing the most analogous story out of aboout 200 candidate stories.

Each task comes in with different lenghts:

S1:: each story is 1-sentence… See the full description on the dataset page: https://huggingface.co/datasets/jhu-clsp/AnaloBench.