INDEX

Explanations

physical postures

The neuron detects standalone numeric tokens used for section or list numbering (e.g. headings like “9,” “10,” “11,” etc.).

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

to

-1.54

 from

-1.36

as

-1.18

on

-1.14

||

-1.12

 that

-1.09

for

-1.06

っていると

-1.01

in

-1.00

 this

-1.00

POSITIVE LOGITS

↵↵

1.30

 exemplu

1.07

 conclusione

1.04

Referee

1.03

 restricciones

1.02

其实

1.02

</h2>

1.02

時間は

1.00

Samstag

1.00

其他

0.99

Activations Density 0.000%