INDEX

Explanations

androidclearchroma

This neuron strongly activates on technical tokens such as programming or markup control words, library/class names, file-format directives, hex color codes, and similar code/ config identifiers.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

at

-1.59

and

-1.58

all

-1.57

 other

-1.40

on

-1.38

 some

-1.36

any

-1.31

or

-1.31

 their

-1.30

 with

-1.29

POSITIVE LOGITS

を行い

1.31

笵

1.26

戋

1.25

 vocale

1.24

 lanterna

1.21

 jednej

1.17

ientras

1.16

鹇

1.16

Merkmale

1.16

 ___________

1.13

Activations Density 0.192%