INDEX

Explanations

special characters and diverse scripts

The neuron fires on numeric tokens (digits and numbers) and other standalone symbols representing numerical data.

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

in

-1.37

and

-1.28

on

-1.24

 those

-1.18

as

-1.16

-1.15

 want

-1.13

 Want

-1.10

 your

-1.10

Those

-1.09

POSITIVE LOGITS

 volition

1.12

 炒め

1.12

 pił

1.07

取り組

1.06

MÁ

1.05

 temporaire

1.05

 boos

1.05

 که

1.04

PANY

1.03

 contraception

1.03

Activations Density 0.002%