INDEX

Explanations

geo prefixes

The neuron fires strongly on the tokens comprising the name “Gehrig,” i.e. when the document mentions “Gehrig.”

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 לח

0.64

 drag

0.58

">⁢</

0.58

 drags

0.57

ስተ

0.56

 nekaj

0.56

)"))

0.55

 кор

0.55

 పో

0.55

nid

0.54

POSITIVE LOGITS

offrey

0.83

ge

0.82

Ge

0.75

Ge

0.73

ellschaft

0.72

 Geoff

0.69

ometrical

0.68

hegan

0.67

orgetown

0.64

orge

0.63

Activations Density 0.197%