INDEX

Explanations

73

The neuron fires strongly on any token that begins with the letter “g,” including both single “g” tokens and longer tokens starting with “g.”

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

 बंगाल

0.59

ള്ള

0.59

 શુભેચ્છાઓ

0.57

uckle

0.56

स्टैंड

0.55

immers

0.55

ocell

0.55

 يعطيكم

0.55

 reale

0.55

᱖

0.54

POSITIVE LOGITS

co

0.54

dry

0.53

Co

0.52

נן

0.50

ron

0.49

dry

0.49

Hun

0.49

bor

0.48

σεις

0.48

Dry

0.48

Activations Density 0.020%