INDEX

Explanations

programming keywords and scope delimiters

The neuron fires on tokens used in code‐comment directives or annotations that disable or suppress warnings (e.g. “eslint‐disable,” “pylint:disable,” “Suppress(…),” etc.).

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

がて

-1.23

のもの

-1.21

感は

-1.15

ところに

-1.14

mountable

-1.13

:{

-1.08

すっ

-1.08

んですよね

-1.05

感が

-1.02

!!!"

-1.02

POSITIVE LOGITS

がありますが

1.38

もあって

1.30

unno

1.24

による

1.20

again

1.17

因为

1.14

zzet

1.13

ḿ

1.09

uride

1.09

eamiento

1.09

Activations Density 0.009%