News for hackers. Stuff that matters.
Re:Liquid Foundation Models: Our First Series of Generative ... by paradite on Monday September 30, 2024 @02:22PM (#41700332)
If you look at the main image of the MMLU-Pro benchmark comparisons, and compare it to the tables below, you can see how they cherry picked one single benchmark where the models better than all the others.

Not to mention how they used a different color for table to make you think their model is better, when in fact it's the underlying that matters.

Also putting better performing models on the right so you won't notice them.

Very sketchy post all around despite the professional look.

Re:Liquid Foundation Models: Our First Series of Generative ... by causal on Monday September 30, 2024 @02:37PM (#41700481)
And the lack of detail on the architecture innovations are also... odd? Like why make such a big deal about how novel your architecture is and then just explain it as "derived from a set of computational units." ?? okay
Re:Liquid Foundation Models: Our First Series of Generative ... by maxc99 on Monday September 30, 2024 @03:17PM (#41700923)
They explain their architecture in more detail here: https://www.liquid.ai/blog/liquid-neural-networks-research.