INDEX

Explanations

listing additional items

The neuron is looking for the “-ing” suffix on verbs—that is, present-participle/gerund forms split across subword tokens.

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

 only

-2.63

not

-1.54

when

-1.50

 although

-1.48

 sebagainya

-1.43

 while

-1.42

 just

-1.40

just

-1.37

in

-1.36

 mostly

-1.36

POSITIVE LOGITS

reportWebVitals

1.61

牁

1.60

 zonder

1.59

 anderen

1.52

rolex

1.51

genen

1.48

():

1.48

 other

1.45

洐

1.45

และ

1.41

Activations Density 0.091%