INDEX

Explanations

questions starting with how or do

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

itag

1.43

etre

1.10

Objet

1.07

epoca

1.07

objet

1.05

iformes

1.04

uze

1.03

isyon

1.02

amatan

0.97

tien

0.96

POSITIVE LOGITS

 does

4.62

do

4.00

did

3.99

does

3.66

 Does

3.52

 DOES

3.28

Does

3.27

 doth

3.10

did

3.02

are

2.93

Activations Density 0.532%