INDEX

Explanations

With followed by description

The neuron is primarily detecting the sentence‐initial preposition “with” (i.e. clauses introduced by “With”).

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

 when

-1.84

 would

-1.59

 while

-1.51

or

-1.45

and

-1.41

 tevreden

-1.30

when

-1.25

 sehr

-1.23

 which

-1.21

 what

-1.20

POSITIVE LOGITS

1.22

 parfums

1.14

 raiſ

1.13

werten

1.09

¼

1.08

 процесс

1.08

lendir

1.06

ㄲ

1.06

 пищи

1.06

 nouveaux

1.05

Activations Density 0.031%