INDEX

Explanations

attorney and legal briefs

The neuron fires on phrasing that introduces actors’ characters—particularly the “in the lead roles”/cast‐listing construction.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

by

-1.93

for

-1.48

 και

-1.42

 that

-1.38

 such

-1.36

an

-1.34

 only

-1.33

 data

-1.28

 using

-1.28

on

-1.24

POSITIVE LOGITS

cây

1.35

Maks

1.35

 ホルダー

1.31

 küm

1.30

 erhi

1.28

ケモン

1.27

 discuter

1.27

 élar

1.26

Mange

1.25

 réjou

1.23

Activations Density 0.004%