INDEX

Explanations

place, with, difference, works, gone

The neuron flags procedural or instructional cue phrases—brief function words (like “in,” “to,” “following”) when they introduce steps or directions.

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

 different

-1.27

 Both

-1.23

お気軽

-1.21

户

-1.20

[*

-1.14

 муніципалі

-1.13

 Likewise

-1.13

ई

-1.13

--"

-1.13

畅

-1.11

POSITIVE LOGITS

 when

1.30

as

1.25

 again

1.18

 menggel

1.17

 kostet

1.16

␥

1.15

 BATTERY

1.11

 mengucap

1.09

 vérit

1.09

 once

1.08

Activations Density 0.015%