BoolQ
Harbor adapterReading comprehension dataset that queries for complex, non-factoid information, and require difficult entailment-like inference to solve.
Verified100%Knowledge & Reasoning
- Samples
- 3,270
- Released
- 2019
- Scores
- 162
How to run
ltb · mgsm
$ npm i -g lettokenburn-cli$ ltb bench pull mgsm --n 50$ ltb bench run ./mgsm -a claude-code -m anthropic/claude-opus-4-8$ ltb bench push
Runs happen on your machine with your own subscription; only the score summary is pushed. CLI docs
Scores
6 sourced entries| Harness | Source | Confidence | Link | |||
|---|---|---|---|---|---|---|
| PaLM 2 | 72.2 | N/A | Independent | medium | May 17, 2023 | arxiv.org |
| PaLM 540B | 55.0 | N/A | Independent | medium | Apr 5, 2022 | arxiv.org |
| text-davinci-003 | 36.0 | N/A | Independent | medium | Oct 20, 2022 | arxiv.org |
| code-davinci-002 | 35.0 | N/A | Independent | medium | Oct 20, 2022 | arxiv.org |
| text-davinci-002 | 23.7 | N/A | Independent | medium | Oct 20, 2022 | arxiv.org |
| GPT-3 Davinci 175B | 5.7 | N/A | Independent | medium | Oct 20, 2022 | arxiv.org |
Every number links to where it was published. Low-confidence rows are muted and need per-cell verification.
Reproduced by the community
All runsLoading community runs…
About
MGSM (Multilingual GSM) — 多语言数学应用题. Scoring: 准确率. Difficulty: 高. Imported from benchmark_research.md (2026-06).