INDEX

Explanations

inserting things into places

The neuron fires strongly on verbs and verb forms that describe embedding or inserting something into something else (e.g., plant, implant, inject).

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

 restant

-1.50

 samlet

-1.44



-1.43

 valgt

-1.40

긱

-1.32

 descrição

-1.30

！(

-1.30

 châ

-1.27

 frança

-1.25

看着

-1.25

POSITIVE LOGITS

in

1.66

 añade

1.45

了不少

1.45

 into

1.39

他不

1.34

1.31

ftrag

1.30

谄

1.30

by

1.24

cinoma

1.23

Activations Density 0.079%