INDEX

Explanations

digit questions

This neuron fires on words that appear as the final token of a heading or title (e.g. the last word in a section heading).

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

its

-1.49

なのかな

-1.06

 dourada

-1.03

 amarela

-1.01

nitř

-1.01

ﱠ

-1.00

 fisuras

-0.98

他又

-0.96

)]=

-0.94

prar

-0.93

POSITIVE LOGITS

of

1.42

を

1.39

は

1.27

が

1.08

項目

1.04

 culturali

0.93

Altre

0.91

ちなみに

0.91

peutic

0.91

0.89

Activations Density 0.001%