INDEX

Explanations

document game

This neuron fires on non-English tokens—especially Chinese characters, numbers, and related punctuation.

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

你不要

-1.56

了一眼

-1.52

куртка

-1.48

those

-1.45

 تلك

-1.30

终究

-1.28

 ويمكن

-1.27

 eğer

-1.24

 sabias

-1.24

cketed

-1.24

POSITIVE LOGITS

 with

2.08

 each

1.57

one

1.54

all

1.48

 Each

1.43

 this

1.37

 four

1.36

 elettrico

1.36

 first

1.32

 time

1.30

Activations Density 0.028%