(www.zyphra.com)

282 points dataminer | 1 comments | 14 Oct 24 22:45 UTC | HN request time: 0.194s | source

Show context

potatoman22 ◴[14 Oct 24 23:43 UTC] No.41843419[source]▶

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.

replies(1): >>41854593 #

1. hack_ml ◴[16 Oct 24 00:48 UTC] No.41854593[source]▶

>>41843419 #

The ablation studies and the dataset can be found here: https://www.zyphra.com/post/building-zyda-2

↑

Zamba2-7B