INDEX

Explanations

project, forming, quote, installation, rate

The neuron fires on individual terms that name technical tasks, steps, or specialized concepts (e.g. “test,” “project,” “break,” “forming,” “quote,” model numbers, etc.).

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

 bijna

-1.63

 again

-1.50

 všetky

-1.45

ımda

-1.38

等は

-1.38

 better

-1.29

 캐릭터

-1.29



-1.27

 znovu

-1.27

 opnieuw

-1.23

POSITIVE LOGITS

ів

1.46

 dessen

1.36

mathrm

1.31

рец

1.27

if

1.23

 հղումներ

1.15

िल्म

1.15

伊

1.14

waniem

1.12

Activations Density 0.042%