in the loop

phi-3-medium 14B
#194 of 274 scored.

MicrosoftApr 23, 2024Open weights

Capabilities Index

Range 113.8–124.1

121.2

Context

Tokens

—

Input

Per 1M tokens

—

Output

Per 1M tokens

—

Benchmarks

How it scores.

  • GPQA Diamond3.5%

    175th of 186

Everything else Epoch has for it

  • ARC AI288.8%
  • OpenBookQA83.2%
  • HellaSwag76.5%
  • BBH75.2%
  • TriviaQA73.9%
  • MMLU70.7%
  • Winogrande63.0%
  • ANLI33.7%
  • MATH level 517.6%

On the index

Around it.

  1. #191Mixtral 8x22BMistral122.0
  2. #192Mistral LargeMistral · $2.00 in122.0
  3. #193phi-3-small 7.4BMicrosoft121.8
  4. #194phi-3-medium 14BMicrosoft121.2
  5. #195Claude 3 SonnetAnthropic120.7
  6. #196Claude InstantAnthropic120.3
  7. #197Claude 2Anthropic120.1

Sources