Rosinality(@rosinality):https://t.co/uJ3YSeOlhE Recurrent transformer that injects activations from the top layers to the bottom layers at the next step (https://t.co/bubRFyPaBt). Why does this work without training?

註冊並分享邀請連結，可獲得影片播放與邀請獎勵。

立即註冊

Rosinality

@rosinality

ML Engineer @poolsideai

加入 October 2008

1K 正在關注 8K 粉絲

Rosinality@rosinality

2026.08.19 08:11

Recurrent transformer that injects activations from the top layers to the bottom layers at the next step ( Why does this work without training?

顯示更多

0

0

32

1.1K

105

轉發到社區

熱門用戶

4.2M 粉絲

26.4M 粉絲

568.8K 粉絲

オリコンニュース

1.9M 粉絲

吴说区块链

@wublockchain12

181.4K 粉絲

@Kongkongda5882

34.1K 粉絲

ツイッター速報〜BreakingNews

256.7K 粉絲

币安Binance华语

464.5K 粉絲

中國新聞社

547.4K 粉絲

100.8K 粉絲

3.4M 粉絲

348.8K 粉絲

モデルプレス

2.1M 粉絲

一劍浣春秋

231.9K 粉絲

16.8M 粉絲