INDEX

Explanations

may be

The neuron activates on the modal auxiliary “may” (often in “may be” constructions) indicating possibility or permission.

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

踽

-1.41

 Gaulle

-1.31

)}{\

-1.30

ține

-1.30

tij

-1.28

$)$

-1.27

赁

-1.26

wi

-1.25

ága

-1.24

zembro

-1.23

POSITIVE LOGITS

be

1.51

其他

1.51

 erlä

1.48

ついに

1.41

1.40

Oltre

1.39

汌

1.39

However

1.39

 інших

1.30

他の

1.30

Activations Density 0.124%