INDEX

Explanations

Lead Ore, enlightenment Wisdom, Vaadin

The neuron fires on isolated numeric tokens (dates, IDs, counts, version numbers, etc.).

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

できなかった

-0.98

bag

-0.91

)$\\

-0.84

:{

-0.79

gebn

-0.78

ってくる

-0.77

();*/

-0.77

anda

-0.77

that

-0.76

coglie

-0.76

POSITIVE LOGITS

 whatnot

0.92

옹

0.91

Dados

0.90

很

0.90

 Spons

0.86

 regata

0.84

 nyingi

0.82

 πρ

0.81

 Servant

0.81

 aprecia

0.79

Activations Density 0.072%