INDEX

Explanations

code explanations

The main thing this neuron does is detect numeric specification tokens—numbers, measurements, and related colons or units in product descriptions.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

People

0.89

stylesheets

0.89

 えっと

0.86

 Charities

0.82

 関係

0.82

 사람들이

0.81

そういう

0.80

personen

0.80

Mathematics

0.80

ittees

0.80

POSITIVE LOGITS

 unrivalled

1.24

 unrival

1.16

 uncompromising

1.16

 güçlü

1.12

 мощ

1.10

 unparalleled

1.10

 unsurpassed

1.09

 outperforms

1.04

 optimized

1.02

 unmatched

1.00

Activations Density 0.257%