INDEX

Explanations

assist with, relevant to, launched earlier

The neuron fires on prominent keywords that introduce or headline informational sections—words like “assist,” “relevant,” “for,” “my,” etc., marking the start of a list, guide, or outline.

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

and

-1.70

 only

-1.61

but

-1.43

our

-1.34

at

-1.27

 always

-1.23

 such

-1.20

on

-1.20

POSITIVE LOGITS

 vrijdag

1.58

 Перейти

1.56

дек

1.46

riwal

1.45

颯

1.40

 vorige

1.40

蚴

1.39

犼

1.38

vœ

1.38

haltet

1.37

Activations Density 0.440%