INDEX

Explanations

finds or sees something

The neuron activates on past-tense verbs (both regular “-ed” forms and common irregulars).

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

 deoarece

-1.45

 family

-1.42

our

-1.27

 estadía

-1.27

de

-1.27

 potvr

-1.27

 haremos

-1.26

 Children

-1.26

 Women

-1.26

POSITIVE LOGITS

 findes

1.33

Creates

1.32

ઃ

1.31

ppled

1.28

bottomRight

1.28

 Retrieves

1.25

fiés

1.25

нимание

1.23

 romantis

1.22

 unica

1.20

Activations Density 0.026%