INDEX

Explanations

United States Court of Appeals

The neuron consistently fires on multi‐letter tokens that occur at the ends of longer, complex words (i.e. common word endings or suffix fragments).

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

 не

-1.30

Cyfeiriadau

-1.29

 stunning

-1.27

 filosof

-1.25

 pasto

-1.20

ResumeLayout

-1.20

 тоже

-1.19

轧

-1.18

UTELY

-1.18

 スポ

-1.18

POSITIVE LOGITS

to

1.98

it

1.55

 môžu

1.46

 them

1.38

of

1.37

 могли

1.32

ಅ

1.32

at

1.32

but

1.27

>{@

1.24

Activations Density 0.184%