INDEX

Explanations

This neuron fires on the running body text of a news story (common function words, prepositions, and verbs in continuous narrative) rather than on headlines or standalone lines.

locations and events

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

now

-1.25

so

-1.14

lň

-1.10

-1.07

 every

-1.05

 además

-1.04

 each

-1.01

our

-0.99

 such

-0.98

or

-0.98

POSITIVE LOGITS

 ดู

1.20

 voulu

1.17

 وكان

1.16

 firmy

1.11

 emperador

1.10

 aproximado

1.10

この日

1.09

當天

1.08

参见

1.08

 thèse

1.07

Activations Density 0.052%