INDEX

Explanations

proper nouns and titles

The neuron is a “content‐word” detector—it lights up on uncommon, information‐bearing words such as names, acronyms, species names, model numbers, and other rare or specialized terms.

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

 this

-1.32

not

-1.25

 their

-1.19

 without

-1.18

 just

-1.16

or

-1.15

as

-1.14

 brutal

-1.10

 that

-1.09

so

-1.08

POSITIVE LOGITS

zwi

1.23

 erhi

1.14

 COULD

1.13

鸶

1.10

 repuesto

1.07

 gewi

1.05

忘れて

1.05

selbe

1.05

 perpé

1.03

蛲

1.03

Activations Density 0.064%