INDEX

Explanations

so i can

The neuron is keyed to spotting numeric expressions (digits, decimals, counts) in the text.

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

粝

-1.03

KNOW

-0.94

ཎ

-0.93

でない

-0.91

+;

-0.90

をかける

-0.89

extrême

-0.89

わかります

-0.88

ależy

-0.87

AKB

-0.87

POSITIVE LOGITS

 remain

1.26

而是

1.21

 still

1.20

nor

1.14

 rely

1.05

can

1.05

只能

1.05

 continue

1.02

 continues

0.99

it

0.98

Activations Density 0.058%