INDEX

Explanations

process for

This neuron fires most strongly on the formal “process … for” phrasing typical of patent descriptions (e.g. “process for the hydrogenation…”).

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

its

-2.09

 would

-1.46

 either

-1.43

訛

-1.42

 smaller

-1.38

なんだろう

-1.35

 include

-1.30

 Bemerkungen

-1.30

 everything

-1.26

 działalności

-1.26

POSITIVE LOGITS

fär

1.52

 lainnya

1.45

㍑

1.38

scrollPane

1.34

 mentale

1.29

"..."

1.27

Namibia

1.27

 дори

1.27

＇

1.26

konstruktion

1.25

Activations Density 0.092%