INDEX

Explanations

transaction if status updates capabilities cannot final rectal

the neuron activates on single words that serve as key technical or domain-specific nouns (e.g. “transaction,” “VLOOKUP,” “accept,” “heat,” “status”).

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

in

-1.64

at

-1.30

 within

-1.22

 without

-1.20

by

-1.15

 Unlike

-1.14

of

-1.13

)।

-1.13

to

-1.13

 immediately

-1.12

POSITIVE LOGITS

 nettoyer

1.31

 arrêter

1.26

 Dinas

1.15

 enlever

1.13

 mélanger

1.10

 énorm

1.09

 déchets

1.06

 envoyer

1.05

Ă

1.04

 både

1.03

Activations Density 0.080%