Dataset · Hugging Face

localgate/mmlu-pro-closed

Name: mmlu-pro-closed
Creator: localgate
License: mit
Keywords: task_categories:text-classification, task_categories:question-answering, task_ids:multi-class-classification, annotations_creators:machine-generated, language_creators:found, source_datasets:extended|TIGER-Lab/MMLU-Pro, language:en, license:mit, size_categories:10K<n<100K, format:parquet, modality:tabular, modality:text, library:datasets, library:pandas, library:polars

MMLU-Pro Closed — Gemma 4 E2B Success Rates Last updated: 2026-09-07. 12,032 MMLU-Pro questions, each answered five times by google/gemma-4-E2B-it. The dataset records the selected options and success counts under a fixed multiple-choice protocol. These are measurements of one model, not human ratings of question difficulty. This release uses the verified August 12, 2026 rerun. It supersedes the July labels previously published here and applies the study's frozen partition… See the full description on the dataset page: https://huggingface.co/datasets/localgate/mmlu-pro-closed.

View on Hugging FaceReport an issue

Author / organization	localgate
License (as declared)	mit
Downloads	156 (as reported by Hugging Face)
Likes	0
Created	11 August 2026
Last updated at source	7 September 2026

llm-routingdifficulty-estimationtext

Embed this entry

<iframe src="https://silytics.si/embed/item/6b3b109a-56c9-4df8-96f3-2f52a71a0eb4" width="420" height="190" style="border:0" loading="lazy" title="Silytics entry"></iframe>

Free, no tracking. More about feeds and embeds.

More from localgate

localgate/mmlu-pro-openDataset

Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 26 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.