INDEX

Explanations

if I am not wrong

The neuron fires on tokens in confirmation‐of‐understanding phrases (e.g. “I understand correctly,” “If I understand correctly,” etc.).

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

is

-1.76

for

-1.70

in

-1.64

on

-1.52

</h2>

-1.52

-1.48

-1.34

してくれた

-1.20

逡

-1.16

</b>

-1.14

POSITIVE LOGITS

líben

1.48

有很多

1.48

americanos

1.45

 consommer

1.43

 attiv

1.42

 vielä

1.42

.}(

1.41

 bezw

1.40

 बीच

1.38

 récla

1.38

Activations Density 0.026%