INDEX

Explanations

address

The neuron activates on numeric tokens—especially numbers or number‐based sequences (including decimals) such as phone numbers, dates, amounts, or postal codes.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

chromiumfx

0.57

 girlfriends

0.56

 psal

0.55

numericUpDown

0.55

 pigmentation

0.54

ipotent

0.53

 магнит

0.53

 magnetism

0.52

 animais

0.52

 cifra

0.52

POSITIVE LOGITS

 headquarters

1.39

 Address

1.38

Address

1.34

 Headquarters

1.30

address

1.27

 Bldg

1.27

〒

1.27

 ถนน

1.26

 offices

1.24

 address

1.23

Activations Density 0.297%