INDEX

Explanations

nothing left to say or do

The neuron lights up on stop‐word–rich, “end of the road” or rhetorical “nothing more to do/say” constructions (e.g. “nothing left to say or do,” “what else can you do?”).

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

Also

-1.21

ဉ

-1.16

 אלה

-1.13

龶

-1.09

They

-1.06

 minder

-1.06

Those

-1.05

Not

-1.05

刪除

-1.05

»,

-1.04

POSITIVE LOGITS

1.23

 insatiable

1.20

 avaliações

1.16

for

1.10

 disgraced

1.09

 หลัง

1.08

 Viability

1.06

 desirability

1.05

to

1.03

 puisque

1.02

Activations Density 0.043%