INDEX

Explanations

seeing things below

The neuron activates on words that cue the reader to visually inspect or refer to content—e.g. “see,” “look,” “picture,” “below.”

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

Always

-1.25

 також

-1.11

Why

-1.11

 όμως

-1.10

However

-1.10

皆様

-1.10

 zusammen

-1.07

永远

-1.05

Your

-1.02

 вместе

-1.02

POSITIVE LOGITS

our

1.50

by

1.43

we

1.34

my

1.16

 kablo

1.10

 amplio

1.09

 from

1.08

 resalt

1.06

年以上

1.04

basically

1.02

Activations Density 0.095%