INDEX

Explanations

doing things

This neuron specifically detects the verb “do” in its various inflected forms (do, does, did, doing, done).

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

 volontà

-1.55

 říká

-1.55

 sufrió

-1.54

 disfrut

-1.54

ག

-1.49

 trabajó

-1.48

 llegará

-1.47

Elles

-1.47

 nevoie

-1.47

 perdió

-1.45

POSITIVE LOGITS

by

2.11

to

1.71

as

1.59

in

1.50

–

1.47

 have

1.45

 Already

1.44

—

1.42

 while

1.41

already

1.41

Activations Density 0.090%