学習ダイナミクス

書いてみる

関連タグ

#論文 (12,322)
#DFA (105)
#Grokking (2)
#Myhill (1)
#RNN (70)
#S4モデル (2)

Iteration Head: A Mechanistic Study of Chain-of-Thought

Ikemen Mas Kot
9日前

Grokfast: Accelerated Grokking by Amplifying Slow Gradients

Ikemen Mas Kot
2週間前

The Illusion of State in State-Space Models

Ikemen Mas Kot
1か月前