INDEX

Explanations

could

The neuron fires on modal expressions of possibility or capability (especially the word "could").

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

ቍ

-1.53

と言われる

-1.30

总体

-1.27

</h1>

-1.27

ꦧ

-1.26

damento

-1.14

最后由

-1.14

frequently

-1.13

 modif

-1.10

usually

-1.10

POSITIVE LOGITS

be

1.80

 podría

1.45

1.43

 easily

1.41

BIÉN

1.34

龶

1.30

1.28

 pourrait

1.26

龵

1.24

龱

1.23

Activations Density 0.048%