Skip to content

Dataset · cimec

LAMBADA: last word prediction

cimec/lambada

English narratives for testing text comprehension via last word prediction.

What's inside

LAMBADA is a set of English narrative excerpts for the word prediction task. A human guesses the last word when seeing the entire excerpt, but not just the last sentence. The set is used to evaluate text comprehension by language models; size ranges from 10K to 100K examples.

What you can do with it

  • Evaluating causal language models on long-context comprehension
  • Word prediction benchmark
  • Model comparison on LAMBADA
Languages
English
Size
10K–100K rows
Downloads this month
17.5K
Likes
67

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("cimec/lambada")