INDEX

Explanations

concepts followed by states

The neuron fires on abstract topic–setting nouns (like “scene,” “phenomenon,” “process,” “situation,” “story,” “career,” etc.) that label or frame the subject under discussion.

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

of

-1.80

and

-1.48

뎁

-1.37

to

-1.36

 with

-1.18

for

-1.10

 which

-1.09

 from

-1.04

他們

-1.04

 while

-1.02

POSITIVE LOGITS

is

1.71

iramente

1.60

was

1.41

are

1.32

🢒

1.27

preneurship

1.21

 têm

1.20

achts

1.19

 fanno

1.16

ograft

1.16

Activations Density 0.064%