INDEX

Explanations

awarded by or given to

This neuron fires strongly on numerical tokens and numeric values (dates, counts, scores, percentages).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

aucune

-1.41

-1.38

¤

-1.32

aucun

-1.31

లో

-1.27

INCLUD

-1.27

 boda

-1.26

 assurent

-1.25

皞

-1.24

 значительно

-1.23

POSITIVE LOGITS

for

2.13

2.09

an

1.59

on

1.46

one

1.39

 widespread

1.27

at

1.25

 every

1.23

 from

1.22

it

1.21

Activations Density 0.107%