INDEX

Explanations

photo captions

This neuron fires on tokens that appear in image captions and photo credit lines (e.g. “Photo,” “Caption,” numeric dimensions, slashes, photographer/publication names).

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

着一个

-1.52

着一

-1.16

<bos>

-1.15

 Either

-1.13

-1.12

Isn

-1.09

了一口

-1.08

kiksi

-1.08

 only

-1.05

POSITIVE LOGITS

 bahwa

1.47

鑷

1.47

 éstos

1.46

quini

1.41

,'"

1.34

 sommige

1.34

 крупных

1.33

They

1.33

脸上的

1.30

还不错

1.27

Activations Density 0.042%