Dec0ySquid retweetledi
Dec0ySquid
5K posts

Dec0ySquid
@Dec0ySquid
Radiologist. Yoko Taro follows me. Not related to Decoy (Shohei Ohtani’s Dog).
I ain't no La Li Lu Le Lo Katılım Kasım 2018
232 Takip Edilen89 Takipçiler
Dec0ySquid retweetledi

If data centers aren't getting completed, there's no reason to buy NVIDIA GPUs, no reason to buy Sandisk storage, no reason for SK Hynix's stock to keep running, and no foundation underlying the semiconductor boom.
I'm very concerned.
wheresyoured.at/where-are-all-…


English

@AL33012 Measure twice, cut once, still gets it wrong, but it worked anyway~
Must be running a luck build 🤭
English

国際宇宙ステーション #ISS からの電波、三重県津市から液体アンテナでも取れました!
CMの都合で高度の低い18:30頃までの受信でしたが51でしっかり復調しました👍
コップの水位は14センチ位で2m のλ/4には満たないですが波長短縮効果である程度のマッチングにはなっていたと思われます♪
日本語
Dec0ySquid retweetledi
Dec0ySquid retweetledi
Dec0ySquid retweetledi
Dec0ySquid retweetledi
Dec0ySquid retweetledi
Dec0ySquid retweetledi
Dec0ySquid retweetledi
Dec0ySquid retweetledi

"Manifold Steering Reveals the Shared Geometry of Neural Network Representation and Behavior"
Most activation steering treats concepts like directions, but model representations are often curved, so straight line steering can cut through unnatural internal states.
The key idea of the paper is to steer along the activation manifold instead of through it.
Across LLM reasoning tasks and a visual world model, manifold steering produced smoother and more natural behavior, while linear steering caused probability mass or visual states to jump abruptly.

English
Dec0ySquid retweetledi
Dec0ySquid retweetledi
Dec0ySquid retweetledi
Dec0ySquid retweetledi

The ANBERNIC RG ROTATE! 📷
The new Anbernic console with a rotating screen is now available for purchase, will you give it a try?
Link: s.click.aliexpress.com/e/_c4D7z7rv
#anbernic #rgrotate



English
Dec0ySquid retweetledi
Dec0ySquid retweetledi
Dec0ySquid retweetledi





























