INDEX

Explanations

system components and specifications

The neuron fires on words that name measurable quantities or units (e.g. “term,” “rate,” “unit,” “spaces,” “files,” “delivery,” “capital”).

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

Still

-1.24

Heat

-1.15

 apropiado

-1.12

Like

-1.10

 chré

-1.09

erencia

-1.09

 گیاه

-1.06

Although

-1.06

 tuyau

-1.06

اگر

-1.04

POSITIVE LOGITS

to

1.58

ִּ

1.52

for

1.35

ַּ

1.30

וּ

1.27

ּוֹ

1.27

and

1.27

 лука

1.25

 pentru

1.20

 бампер

1.16

Activations Density 0.039%