BoolQ
Progress Over Time
Interactive timeline showing model performance evolution on BoolQ
BoolQ Leaderboard
| Context | Cost | License | ||||
|---|---|---|---|---|---|---|
| 1 |  Nous Research | 70B | — | — | ||
| 2 |  Google | 27B | — | — | ||
| 3 |  Microsoft | 60B | — | — | ||
| 4 | Google | 9B | — | — | ||
| 5 | 2B | — | — | |||
| 5 | Google | 8B | — | — | ||
| 7 | Microsoft | 4B | — | — | ||
| 8 | Microsoft | 4B | — | — | ||
| 9 | 2B | — | — | |||
| 9 | Google | 8B | — | — |
What is BoolQ?
BoolQ is a reading comprehension dataset for yes/no questions containing 15,942 naturally occurring examples. Each example consists of a question, passage, and boolean answer, where questions are generated in unprompted and unconstrained settings. The dataset challenges models with complex, non-factoid information requiring entailment-like inference to solve.
BoolQ is a text benchmark evaluating models on reasoning and language tasks. LLM Stats tracks 10 models on this benchmark, scored on a 0–1 scale. The current average is 0.8, with the leader at 0.9.
Compare leaders on the best AI for reasoning and best AI for language leaderboards.
Current leaders
Hermes 3 70B from Nous Research currently leads the BoolQ leaderboard with a score of 0.880 across 10 evaluated AI models.
Source paper
- Title
- BoolQ: Exploring the Surprising Difficulty of Natural Yes/No Questions
- Authors
- Christopher Clark, Kenton Lee, Ming-Wei Chang, Tom Kwiatkowski, and 2 others
- Published
- arXiv
- 1905.10044
Abstract
In this paper we study yes/no questions that are naturally occurring --- meaning that they are generated in unprompted and unconstrained settings. We build a reading comprehension dataset, BoolQ, of such questions, and show that they are unexpectedly challenging. They often query for complex, non-factoid information, and require difficult entailment-like inference to solve. We also explore the effectiveness of a range of transfer learning baselines. We find that transferring from entailment data is more effective than transferring from paraphrase or extractive QA data, and that it, surprisingly, continues to be very beneficial even when starting from massive pre-trained language models such as BERT. Our best method trains BERT on MultiNLI and then re-trains it on our train set. It achieves 80.4% accuracy compared to 90% accuracy of human annotators (and 62% majority-baseline), leaving a significant gap for future work.
FAQ
Common questions about the BoolQ benchmark and leaderboard.