INDEX

Explanations

Starting words of lists and phrases

The neuron is looking for large standalone numeric tokens (e.g. four‐digit metadata numbers).

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

as

-1.29

-1.13

 whose

-1.12

 with

-1.06

on

-1.04

all

-1.04

if

-1.04

 even

-1.02

now

-1.02

POSITIVE LOGITS

 immediatamente

1.21

ず

1.20

Tils

1.19

 illuminating

1.19

いろんな

1.18

 refuerzo

1.18

푉

1.16

 ağaç

1.14

;</

1.13

AddHtmlAttribute

1.13

Activations Density 0.140%