Skip to content

google/bigbench

Multiple ChoiceQuestion AnsweringText ClassificationText GenerationZero Shot ClassificationOtherENBenchmarkapache-2.0

Google/bigbench is a multiple choice-focused benchmark dataset in EN that provides 5,262,494 labeled examples distributed in Parquet format. It is distributed under the apache-2.0 license, and has been downloaded 491 times.

📊 This dataset is used as an LLM benchmark. See model leaderboards →

About google/bigbench

The Beyond the Imitation Game Benchmark (BIG-bench) is a collaborative benchmark intended to probe large language models, and extrapolate their future capabilities.

Details

Task
Multiple Choice, Question Answering, Text Classification, Text Generation, Zero Shot Classification, Other
Language
EN
Format
Parquet
Rows / instances
5262494
Creator
google
Year
2022
License
apache-2.0
Downloads
491
Likes
72
Download Homepage

FAQ