INDEX

Explanations

and then action verbs

The neuron fires on the second verb in “to X and Y”‐style coordinated verb phrases (e.g. “to introduce and analyze,” “to create and evaluate”).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

？』

-1.29

但却

-1.27

を

-1.25

もしか

-1.22

などの

-1.16

 other

-1.14

曜日

-1.14

ですが

-1.11

<bos>

-1.09

憩

-1.07

POSITIVE LOGITS

為に

1.51

 then

1.39

ようです

1.29

様な

1.26

 становятся

1.21

ようですが

1.14

ようで

1.10

conceito

1.09

ナメント

1.06

fazer

1.05

Activations Density 0.043%