INDEX

Explanations

and conjunctions

The neuron selectively activates on long hyphenated compound tokens (e.g. multi‐word slugs or file/URL fragments).

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

𝙳

0.44

ُ

0.44

 FontFamily

0.43

 Enseñ

0.43

 Гу

0.42

 Gson

0.40

 JavaBeans

0.40

\_

0.40

𝙶

0.40

 สืบค้นเมื่อ

0.39

POSITIVE LOGITS

এর

0.44

和大

0.42

and

0.42

for

0.41

打造

0.41

 येतो

0.40

ttes

0.38

with

0.38

可以

0.38

 unites

0.38

Activations Density 0.000%