INDEX

Explanations

hexadecimal bitwise operations

The neuron fires on tokens that are numeric (digits, decimal values, big numbers). It’s essentially detecting numbers/quantitative figures.

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

ни

1.16

с

1.12

ل

1.09

iyorum

1.05

leicht

1.02

 planas

1.02

го

1.01

вью

1.01

 نقدم

1.01

errorhandler

1.00

POSITIVE LOGITS

년대

1.16

ية

1.13

ات

1.13

 arcane

1.03

ﻟ

1.01

ان

1.00

 obscurity

1.00

}}

0.97

Osm

0.95

ti

0.88

Activations Density 2.840%