Will an open source model beat any gpt-4 in eq-bench by the end of February 2024?
Mini
32
3.2k
resolved Mar 6
Resolved
YES

I will benchmark it myself.

Here, open source means weights publicly available regardless of license.

Get Ṁ600 play money

🏅 Top traders

#NameTotal profit
1Ṁ75
2Ṁ31
3Ṁ30
4Ṁ19
5Ṁ13
Sort by:

ShinojiResearch/Senku-70B-Full scores higher on EQ-Bench than gpt-4-0613 and gpt-4-0125-preview as per eqbench.com

should a model based on leaked mistral weights count as open? Market description seems to imply yes.

Can someone benchmark this https://huggingface.co/axra/miqu-120b