INDEX

Explanations

citations with numbers and letters

The neuron fires on the little single‐character markers that appear in citations and headings—standalone digits, asterisks, and the “a”/“b” suffixes used to distinguish reference entries.

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

 qualche

-1.34

 filós

-1.24

ſam

-1.20

itern

-1.15

を実行

-1.12

の準備

-1.11

へ行

-1.06

麁

-1.04

いくつか

-1.03

 предотвра

-1.02

POSITIVE LOGITS

谷川

1.27

1.05

是個

1.02

餃

1.00

思います

1.00

時間は

0.98

しかし

0.98

けどね

0.98

(!)

0.97

は

0.97

Activations Density 0.004%