INDEX

Explanations

colour selection

The neuron activates on appearances of the word “colour” (and its close variants) in the text.

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

 быстро

-2.25

贗

-2.22

sandalia

-2.19

 Hälfte

-2.16

 詳細

-2.16

轻声

-2.13

 Familienname

-2.09

 ซึ่ง

-2.08

指着

-2.05



-2.05

POSITIVE LOGITS

呖

2.94

 και

2.70

榈

2.59

we

2.52

și

2.34

圌

2.34

and

2.33

 подой

2.31

 publicized

2.30

––

2.25

Activations Density 0.012%