INDEX

Explanations

hill and hall

The neuron fires on mentions of the “Hall‐effect” (and closely related terms like Hall cell and hall‐effect sensor).

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

↵↵

-2.03

 TOPOLOGY

-1.52

of

-1.45

し

-1.45

ва

-1.45

з

-1.43

\"

-1.41

н

-1.41

玧

-1.36

oplayer

-1.34

POSITIVE LOGITS

 podrían

1.75

 estadía

1.62

 eenige

1.58

PUESTA

1.55

 soooo

1.51

鼍

1.51

蛲

1.51

 adaptación

1.50

铊

1.50

няется

1.48

Activations Density 0.012%