INDEX

Explanations

first or second instance

This neuron detects “component–of–whole” noun phrases—i.e. constructions of the form “X of the Y,” like “part of the event,” “suite of the complex,” or “western part of the settlement.”

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

 både

-1.56

 allerlei

-1.37

ulante

-1.25

😕

-1.21

 zowel

-1.21

 pittores

-1.17

 député

-1.15

 tunik

-1.12

 véritable

-1.12

Ꮚ

-1.09

POSITIVE LOGITS

 first

1.65

one

1.57

 even

1.30

 only

1.30

 last

1.29

 εισ

1.24

 second

1.24

 there

1.20

 later

1.19

 main

1.13

Activations Density 0.073%