INDEX

Explanations

fluid pump

The neuron fires on unusually long, multi‐subword tokens—i.e. complex or highly specialized terms that span several word‐pieces.

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

 will

-1.73

-1.48

).

-1.36

–

-1.34

--

-1.30

—

-1.30

-1.19

',

-1.18

own

-1.15

POSITIVE LOGITS

ELECTRIC

1.29

 furthermore

1.28

 sekitar

1.27

 simply

1.25

 ONLY

1.24

 specifically

1.23

1.22

 केवल

1.21

 simplement

1.20

Unternehmen

1.19

Activations Density 0.234%