INDEX

Explanations

instructions for or invitations to

The neuron strongly activates on the elements of procedural or instructional text—especially list-item numbers and the accompanying action words (e.g. “can,” “contribute,” “using,” “to,” etc.) that mark steps.

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

 even

-2.03

so

-1.82

 also

-1.80

 other

-1.61

 many

-1.49

if

-1.48

felizmente

-1.45

 additional

-1.37

 didn

-1.34

it

-1.34

POSITIVE LOGITS

Either

1.41

 câte

1.30

傒

1.30

 segítség

1.27



1.26

 jednod

1.25

ሾ

1.24

 gustado

1.24

 ſou

1.23

voyez

1.22

Activations Density 0.115%