INDEX

Explanations

place names followed by descriptive words

The neuron fires strongly on tokens representing numerical values (often written as comma-prefixed numbers), so it’s detecting numeric data in the text.

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

at

-2.33

and

-1.63

for

-1.54

now

-1.43

all

-1.34

up

-1.34

on

-1.34

 newest

-1.34

 that

-1.34

 from

-1.32

POSITIVE LOGITS

🔸

1.60

opies

1.50

 musul

1.50

อีก

1.48

another

1.47

</h1>

1.45

 medie

1.44

𓐍

1.37

鵙

1.37

콴

1.36

Activations Density 0.186%