INDEX

Explanations

The neuron fires strongly on words that begin sentences or clauses—especially sentence‐initial function words like conjunctions (“But,” “It’s”) and pronouns.

their, they, partners

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

 often

-1.52

 always

-1.34

can

-1.34

 still

-1.28

 while

-1.27

 sometimes

-1.23

 even

-1.20

 generally

-1.20

 many

-1.17

so

-1.15

POSITIVE LOGITS

 speciale

1.13

 tanam

1.11

 malen

1.09

 ventre

1.05

 необы

1.05

 applau

1.04

 Meksi

1.02

 cioc

1.01

danke

1.00

Samen

1.00

Activations Density 0.110%