INDEX

Explanations

describing something

The neuron strongly activates on the word “Something,” signaling occurrences of that exact discourse token in the text.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

op

1.73

uding

1.61

é

1.45

š

1.45

1.41

ó

1.34

ô

1.30

rå

1.27

 demás

1.27

ng

1.27

POSITIVE LOGITS

 الشيء

1.26

 defray

1.25

ки

1.24

שהו

1.24

 akin

1.11

 beak

1.09

 impede

1.08

बी

1.07

 immensely

1.05

 $\{(

1.05

Activations Density 0.072%