Dataset · google
Multitask benchmark for language models
google/bigbench
A collection of diverse tasks for evaluating and comparing the capabilities of large language models.
What's inside
This is a collaborative benchmark consisting of multiple tasks for testing large language models and assessing their future capabilities. Metadata specifies multiple-choice, question-answering, text-classification, text-generation, and zero-shot-classification. The language is English.
What you can do with it
- Evaluating LLMs in zero-shot settings across multiple tasks
- Testing models on question-answering and multiple-choice
- Comparing text-generation and text-classification
- Tasks
- Multiple choice, Question answering, Text classification, Text generation, Zero shot classification, Other
- Languages
- English
- Downloads this month
- 558
- Likes
- 74
Models trained on it
How to load it
from datasets import load_dataset
ds = load_dataset("google/bigbench")