INDEX

Explanations

type followed by designation

The neuron activates on occurrences of the word “Type” followed by a classification label (e.g. “Type I,” “Type II,” “Type III,” etc.).

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

 been

-1.46

 from

-1.42

 about

-1.34

 have

-1.24

-1.23

has

-1.22

 also

-1.20

He

-1.16

гинал

-1.14

did

-1.09

POSITIVE LOGITS

𝒾

1.43

 quando

1.41

 когато

1.38

튿

1.32

 когда

1.30

と呼ばれる

1.27

 parku

1.25

荚

1.22

情况下

1.20

quando

1.20

Activations Density 0.164%