INDEX

Explanations

operating systems and services

This neuron flags rare or domain-specific subword tokens (e.g. proper nouns, technical terms, and infrequent fragments).

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

who

-1.80

 that

-1.63

 their

-1.55

any

-1.40

鰊

-1.38

 préférable

-1.33

 veck

-1.30

鬭

-1.29

 kimdir

-1.28

 cotone

-1.27

POSITIVE LOGITS

rightarrow

1.48

tiously

1.34

is

1.33

&#

1.30

vgili

1.28

hetically

1.26

 если

1.24

gered

1.24

攸

1.23

oted

1.21

Activations Density 0.170%