HackerTrans
トップ新着トレンドコメント過去質問紹介求人

foldl2022

no profile record

コメント

foldl2022
·先月·議論
Gemma-4 E2B/E4B models reuses K-V cache from other layers, which do things in a "transposed" way: not reuse Q/K/V matrices within a single layer, but reuse across different layers.
foldl2022
·9 か月前·議論
So, show me the model weights, please.
foldl2022
·10 か月前·議論
A nice side-effect of "variables at the top": you keep your functions short.
foldl2022
·2 年前·議論
Off the topic: I am also in favor of minimalist.

In the case of AlphaGeometry, I made AlphaGeometryRE to get rid of tensorflow/jax/flax, and 100+ Python packages.