INDEX

Explanations

name and context

This neuron strongly activates on the literal word “name,” i.e. mentions of “name” in the text.

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

财

-1.52

咘

-1.41

 rekom

-1.41

 länder

-1.39

们

-1.38

amanca

-1.38

 幸福

-1.37

𝙑

-1.35

 stör

-1.33

 fraî

-1.30

POSITIVE LOGITS

着一个

1.53

1.32

os

1.28

on

1.27

 obtains

1.20

ve

1.16

ot

1.15

 obstru

1.15

 zejména

1.14

Activations Density 0.031%