INDEX

Explanations

swap or exchange

The neuron activates on any form of the root “swap” (e.g. “swap,” “swapped,” “swapping”).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

-1.98

-1.63

-1.39

<em>

-1.38

Ã

-1.36

 الخم

-1.28

 зиму

-1.27

 horse

-1.27

Â

-1.27

ський

-1.26

POSITIVE LOGITS

the

1.83

 between

1.66

if

1.55

čných

1.47

llavero

1.46

without

1.42

 febbraio

1.42

with

1.40

焄

1.39

when

1.38

Activations Density 0.036%