MMLU: College Physics

MMLU subject subset: Undergraduate mechanics, electromagnetism and thermodynamics, typically calculation-heavy.

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
Subcategoryphysics
Page statusactive
Metricaccuracy
Directionhigher_is_better
Unit%
Dataset size102
Dataset licenceMIT
PublisherUC Berkeley (original); Center for AI Safety (current host)

What it measures

Undergraduate mechanics, electromagnetism and thermodynamics, typically calculation-heavy. Questions are four-option multiple-choice, drawn from the MMLU test set's "physics" 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
Meta Llama 3 70BMeta59.82024-07
Meta Llama 3 70B InstructMeta59.82026-04
Meta Llama 3 70B InstructNous Research59.82026-04
Yi 1.5 34B 32K01.AI59.82024-07
Mixtral 8x22B Instruct v0.1Mistral AI57.82026-04
Yi 1.5 34B01.AI56.92026-04
Yi 34B 200K01.AI54.92024-07
Nous Hermes 2 Yi 34BNous Research52.92024-07
Yi 1.5 34B Chat01.AI52.02026-04
Yi 1.5 34B Chat 16K01.AI52.02024-07
Meta Llama 3 8B InstructMeta50.02024-07
Meta Llama 3 8B InstructNous Research50.02024-07
Meta Llama 3 8BMeta49.02024-07
Meta Llama 3 8BNous Research49.02024-07
Nous Hermes 2 Mixtral 8x7B DPONous Research49.02024-07
Hermes 2 Theta Llama 3 8BNous Research48.02024-07
Yi 1.5 9B01.AI47.12024-07
Yi 1.5 9B 32K01.AI47.12024-07
Yi 1.5 9B Chat 16K01.AI47.12024-07
Hermes 2 Pro Llama 3 8BNous Research46.12024-07
Mixtral 8x7B v0.1Mistral AI46.12026-04
Yi 1.5 9B Chat01.AI46.12024-07
Yi 34B Chat01.AI46.12024-07
Yi 1.5 6B Chat01.AI45.12024-07
Yi 1.5 6B01.AI44.12024-07
Mixtral 8x7B Instruct v0.1Mistral AI43.12026-04
Phi 3 mini 128K instructMicrosoft43.12024-07
Mistral 7B Instruct v0.2Mistral AI42.22024-07
Yi 9B01.AI42.22024-07
Nous Hermes 2 SOLAR 10.7BNous Research41.22024-07
gemma 7B itGoogle DeepMind40.22024-07
Qwen2 1.5B InstructAlibaba / Qwen Team40.22024-07
phi 2Microsoft37.32024-07
Phi 3 mini 4K instructMicrosoft36.32024-07
Yi 6B01.AI34.32024-07
Yi 6B Chat01.AI34.32024-07
Mistral 7B v0.3Mistral AI31.42024-07
mistral 7B v0.3 bnb 4bitUnsloth31.42024-07
deepseek llm 7B baseDeepSeek29.42024-07
deepseek llm 7B chatDeepSeek29.42024-07
falcon 40BTII27.52024-07
deepseek coder 1.3B baseDeepSeek25.52024-07
deepseek coder 1.3B instructDeepSeek25.52024-07
deepseek coder 6.7B baseDeepSeek25.52024-07
Qwen2 0.5B InstructAlibaba / Qwen Team24.52024-07
deepseek coder 6.7B instructDeepSeek23.52024-07
OLMo 1B hfAllen AI22.52024-07
chatglm2 6BZhipu AI21.62024-07
gemma 2B itGoogle DeepMind18.62024-07
gemma 2BGoogle DeepMind14.72024-07

Data

This page as JSON · Edit on GitHub