Introducing DBRX: A New State-of-the-Art Open LLM by Databricks

Re:Introducing DBRX: A New State-of-the-Art Open LLM by Data... by n2d4 on Wednesday March 27, 2024 @02:14PM (#39842666)

There's at least an argument to be made that this is because all the models are heavily trained on GPT-4 outputs (or whatever the SOTA happens to be during training). All those models are, in a way, a product of inbreeding.

[ Parent | on HN ]

Re:Introducing DBRX: A New State-of-the-Art Open LLM by Data... by fragmede on Wednesday March 27, 2024 @03:58PM (#39843856)

But is it the kind of inbreeding that gets you Downs, or the kwisatz haderach?

[ Parent | on HN ]

Re:Introducing DBRX: A New State-of-the-Art Open LLM by Data... by batshit_beaver on Wednesday March 27, 2024 @09:17PM (#39846705)

Yes

[ Parent | on HN ]

Re:Introducing DBRX: A New State-of-the-Art Open LLM by Data... by pram on Wednesday March 27, 2024 @02:21PM (#39842753)

Consider the bulldog: https://youtube.com/watch?v=hUgmkCgMWbg

[ Parent | on HN ]

Re:Introducing DBRX: A New State-of-the-Art Open LLM by Data... by sumo43 on Wednesday March 27, 2024 @02:46PM (#39843063)

Maybe true for instruct, but pretraining datasets do not usually contain GPT-4 outputs. So the base model does not rely on GPT-4 in any way.

[ Parent | on HN ]