INDEX

Explanations

methodology and context prepositions

The neuron fires strongly on tokens related to describing experimental methods and quantitative measurements (e.g. “investigate,” “using,” and numerical measurement contexts).

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

 that

-1.70

 kommenden

-1.64

 Instead

-1.52

 должна

-1.48

 would

-1.46

StoryboardSegue

-1.45

 will

-1.43

so

-1.35

to

-1.35

那么

-1.34

POSITIVE LOGITS

 whether

1.66

њи

1.53

by

1.52

1.39

૬

1.38

1.34

cestors

1.31

quels

1.31

 leur

1.30

 oiseau

1.29

Activations Density 0.468%