INDEX

Explanations

use to perform task

This neuron detects imperative instruction phrases that begin with “Use this.”

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

啬

-1.45

alties

-1.42

hibitions

-1.40

willReturn

-1.40

Tage

-1.34

 handen

-1.32

 affaires

-1.32

讓我

-1.31

requirements

-1.30

Anda

-1.25

POSITIVE LOGITS

to

2.05

 only

1.50

 instead

1.45

 тільки

1.41

 this

1.39

 únicamente

1.36

only

1.21

 raczej

1.20

of

1.20

澍

1.19

Activations Density 0.080%