Gary Marcus

Rank 11 of 47

|

Score 83

--

@notdjkhaled7 some don’t have immediately obvious leaderboards, but I took MMLU GPT-2 and 3 from Hendrycks’ site, Turbo from OAI’s new site.

I would love it if you wanted to put it all together including 2, 3, 3.5, 4 and Turbo and perhaps competitors across all the measures just reported.

4/12/2024, 1:32:45 AM

In reply to:

(not dj) khaled

@notdjkhaled7

·

661d

@GaryMarcus They have all what you asked for in the GitHub repo

Gary Marcus

@GaryMarcus

·

661d

Could we see GPT 3 and 3.5 and GPT 4 on the same plot? And Gemini Pro 1.5 and Claude Opus?

OpenAI

@OpenAI

·

661d

Our new GPT-4 Turbo is now available to paid ChatGPT users. We’ve improved capabilities in writing, math, logical reasoning, and coding.
Source: https://github.com/openai/simple…

The statement is a technical request for the aggregation of performance data on various AI models, aimed at enhancing the understanding of their capabilities.