INDEX

Explanations

closing parentheses and brackets

The neuron strongly activates on the closing square‐bracket token and nearby numeric or “for […]” tokens—that is, it detects content inside square brackets (e.g. bracketed citations or numeric asides).

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

 сожалению

-1.33

and

-1.32

 dikel

-1.25

to

-1.24

 نهر

-1.23

 sportivo

-1.23

ul

-1.21

it

-1.20

美しい

-1.20

ed

-1.19

POSITIVE LOGITS

innen

1.31

guitarra

1.30

tische

1.28

ꞈ

1.27

鹚

1.24

aları

1.23

 とき

1.23

dámské

1.23

 のか

1.23

dle

1.22

Activations Density 0.014%