INDEX

Explanations

The neuron lights up on descriptive “content” words—especially adjectives and adverbs—rather than on function words.

manner or degree

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

昨日の

-1.55

際に

-1.48

うわ

-1.37

昨日は

-1.36

に戻る

-1.36

and

-1.32

ままで

-1.30

で見る

-1.28

 desmotiv

-1.24

明日は

-1.23

POSITIVE LOGITS

in

1.61

on

1.48

 into

1.41

 than

1.38

 особенно

1.33

án

1.33

を

1.31

と

1.30

around

1.30

では

1.29

Activations Density 0.032%