News for hackers. Stuff that matters.

A comment on Zamba2-7B.

Re:Zamba2-7B by potatoman22 on Monday October 14, 2024 @07:43PM (#41843419)
I wonder how much of the performance gains can be attributed to their improved dataset rather than their architecture. That would be an expensive experiment.
[ Parent | on HN ]
Re:Zamba2-7B by hack_ml on Tuesday October 15, 2024 @08:48PM (#41854593)
The ablation studies and the dataset can be found here: https://www.zyphra.com/post/building-zyda-2
[ Parent | on HN ]