INDEX

Explanations

formal, most, more, focus

This neuron fires on non‐word punctuation tokens (e.g. ellipses, dashes, quotes, brackets).

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

्स

1.40

as

1.35

ibility

1.35

ian

1.34

ates

1.34

 chords

1.33

ם

1.31

 reflectors

1.30

 refers

1.27

 ghosts

1.27

POSITIVE LOGITS

页面存档备份

1.47



1.42

promo

1.33

perturb

1.30

㧍

1.28

πλ

1.27

dbThoughtData

1.27

θ

1.27

*}$

1.26

英語版

1.25

Activations Density 1.914%