INDEX

Explanations

contact or choice

data flowing and exploiting

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

to

0.49

kamer

0.47

ocarbon

0.44

proposal

0.43

toe

0.43

password

0.43

asiti

0.43

killer

0.42

↵

0.42

prest

0.41

POSITIVE LOGITS

HUN

0.54

Huz

0.54

Nf

0.51

 humides

0.51

 grie

0.51

 Epoch

0.50

 educate

0.50

 FormControl

0.50

Hu

0.49

 CONST

0.48

Activations Density 0.001%