INDEX

Explanations

to make a choice or take action

The neuron fires on strong action verbs used to give instructions, advice, or calls to action (e.g. “read,” “interrogate,” “make,” “shoot,” “question”).

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

 vorrei

-1.42

translates

-1.33

batal

-1.29

or

-1.27

 rác

-1.25

艽

-1.25

geç

-1.19

 unless

-1.19

 passend

-1.18

thétique

-1.17

POSITIVE LOGITS

 people

1.59

 also

1.58

ポーチ

1.30

 INFORMACIÓN

1.30

 边框

1.30

 terbesar

1.30

 valentino

1.27

umi

1.27

 richtige

1.27

>');

1.24

Activations Density 0.028%