INDEX

Explanations

potential and possibility

The neuron fires strongly on nouns naming conditions, resources, or prerequisites (e.g. “circumstances,” “cards,” “specs,” “research,” “luck,” “talent”)—i.e. key factors in advice or how-to statements.

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

minated

-1.17

 poft

-1.01

ificación

-0.98

mability

-0.95

icidio

-0.93

 باید

-0.91

必须

-0.91

nellement

-0.90

ニック

-0.90

灕

-0.90

POSITIVE LOGITS

and

2.03

can

2.02

 will

1.52

 could

1.50

สามารถ

1.28

 можно

1.22

مكن

1.22

 then

1.20

you

1.18

Then

1.17

Activations Density 0.040%