INDEX

Explanations

Names followed by initials

The neuron fires on occurrences of a single uppercase letter followed by a period (i.e. name initials).

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

 seng

-1.32

-*

-1.31

-1.28

</em>

-1.22

-1.20

 warung

-1.19

 tuta

-1.17

 luks

-1.16

USTIN

-1.16

薆

-1.14

POSITIVE LOGITS

by

1.55

or

1.40

 that

1.38

 solchen

1.37

at

1.30

rozco

1.25

 finnas

1.24

 during

1.23

しております

1.21

 from

1.20

Activations Density 0.030%