INDEX

Explanations

The neuron activates on past-participle verb forms (typically ending in “-ed”) indicating completed actions or states.

actions completed and their results

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

 problemet

-1.59

-1.44

کند

-1.41

-1.36

-1.35

 permita

-1.34

-1.21

жень

-1.20

 tenha

-1.19

FF

-1.18

POSITIVE LOGITS

for

1.82

by

1.81

in

1.76

at

1.74

as

1.63

 from

1.60

 with

1.54

detek

1.52

 istimewa

1.52

origami

1.50

Activations Density 0.056%