INDEX

Explanations

learn how to

This neuron detects instructional or tutorial phrasing, especially occurrences of “learn” and “how to.”

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

parseDouble

-1.09

GetInt

-1.02

tellung

-1.00

Karakteristik

-0.99

TextBoxColumn

-0.98

yfikacja

-0.98

bolistas

-0.98

êt

-0.96

CrossRef

-0.95

usitis

-0.95

POSITIVE LOGITS

how

2.80

to

2.64

 about

2.30

 from

2.16

new

1.83

howto

1.57

 more

1.34

the

1.25

how

1.24

 through

1.23

Activations Density 0.024%