MMLU: College Mathematics

MMLU subject subset: Calculus, linear algebra and other undergraduate mathematics.

unassessed

This page is a discovery lead. Nobody has yet assessed it against the catalogue contract, so it carries no disposition. Absence of evidence here is not evidence of staleness.
Categoryknowledge
Subcategorymath
Page statusactive
Metricaccuracy
Directionhigher_is_better
Unit%
Dataset size100
Dataset licenceMIT
PublisherUC Berkeley (original); Center for AI Safety (current host)

What it measures

Calculus, linear algebra and other undergraduate mathematics. Questions are four-option multiple-choice, drawn from the MMLU test set's "math" subcategory within the benchmark's "STEM" top-level group, and are graded on the single correct labelled option.

Task format

Four-option multiple-choice questions, graded on the single correct labelled option; commonly evaluated 5-shot, consistent with the rest of MMLU.

Models reporting this benchmark

These figures come from the model cards, which carry one collection date per card and no per-score attribution. They are shown as reported, not as verified evidence.
ModelProviderScoreCard as of
Yi 1.5 34B01.AI59.02026-04
Meta Llama 3 70BMeta58.02024-07
Meta Llama 3 70B InstructMeta58.02026-04
Meta Llama 3 70B InstructNous Research58.02026-04
Yi 1.5 34B 32K01.AI54.02024-07
Yi 1.5 9B01.AI53.02024-07
Yi 1.5 9B 32K01.AI53.02024-07
Yi 1.5 34B Chat01.AI51.02026-04
Yi 1.5 34B Chat 16K01.AI51.02024-07
Mixtral 8x22B Instruct v0.1Mistral AI49.02026-04
Nous Hermes 2 Yi 34BNous Research49.02024-07
Yi 1.5 9B Chat01.AI48.02024-07
Yi 1.5 9B Chat 16K01.AI47.02024-07
Mixtral 8x7B Instruct v0.1Mistral AI46.02026-04
Mixtral 8x7B v0.1Mistral AI46.02026-04
Nous Hermes 2 Mixtral 8x7B DPONous Research43.02024-07
Yi 9B01.AI43.02024-07
Yi 6B01.AI42.02024-07
Yi 6B Chat01.AI42.02024-07
Yi 1.5 6B Chat01.AI41.02024-07
Phi 3 mini 128K instructMicrosoft40.02024-07
Yi 34B 200K01.AI40.02024-07
falcon 40BTII39.02024-07
Meta Llama 3 8B InstructMeta39.02024-07
Meta Llama 3 8B InstructNous Research39.02024-07
Mistral 7B Instruct v0.2Mistral AI39.02024-07
Yi 1.5 6B01.AI39.02024-07
phi 2Microsoft38.02024-07
Mistral 7B v0.3Mistral AI37.02024-07
mistral 7B v0.3 bnb 4bitUnsloth37.02024-07
Nous Hermes 2 SOLAR 10.7BNous Research37.02024-07
gemma 7B itGoogle DeepMind36.02024-07
Hermes 2 Pro Llama 3 8BNous Research36.02024-07
Hermes 2 Theta Llama 3 8BNous Research35.02024-07
Phi 3 mini 4K instructMicrosoft35.02024-07
deepseek coder 6.7B instructDeepSeek34.02024-07
Yi 34B Chat01.AI34.02024-07
Meta Llama 3 8BMeta33.02024-07
Meta Llama 3 8BNous Research33.02024-07
chatglm2 6BZhipu AI32.02024-07
deepseek coder 6.7B baseDeepSeek32.02024-07
Qwen2 1.5B InstructAlibaba / Qwen Team32.02024-07
deepseek llm 7B baseDeepSeek31.02024-07
deepseek llm 7B chatDeepSeek31.02024-07
Qwen2 0.5B InstructAlibaba / Qwen Team31.02024-07
gemma 2BGoogle DeepMind30.02024-07
gemma 2B itGoogle DeepMind26.02024-07
OLMo 1B hfAllen AI26.02024-07
deepseek coder 1.3B baseDeepSeek23.02024-07
deepseek coder 1.3B instructDeepSeek23.02024-07

Data

This page as JSON · Edit on GitHub