INDEX

Explanations

how it works

The neuron activates on explicit technical term names or acronyms (multi‐word or capitalized domain-specific concepts).

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

 مسائل

0.59

 někter

0.55

 รวมถึง

0.54

 dernières

0.54

WithError

0.52

 fisheries

0.51

 neglecting

0.51

eol

0.51

 รวม

0.50

ंख

0.50

POSITIVE LOGITS

原理

1.48

Basically

1.39

 Basically

1.35

仕組み

1.34

Essentially

1.33

 Essentially

1.30

нцип

1.30

 суть

1.29

 작동

1.28

 basically

1.27

Activations Density 0.237%