INDEX

Explanations

Content Delivery Network

The neuron activates on the first element (the modifying word) of common two-word compound or technical terms (e.g. “grab bars,” “cover crops,” “artificial intelligence,” “social media,” “microtransactions,” “operating systems,” “sports drinks”).

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

щий

0.49

 nahi

0.49

 Besitzer

0.48

ని

0.47

őt

0.47

'}$

0.46

fyr

0.46

 Verme

0.46

 endl

0.45

unun

0.44

POSITIVE LOGITS

 각종

0.55

这些

0.52

 এসব

0.51

各类

0.50

 그런

0.50

фай

0.50

 които

0.50

各种

0.49

‐

0.47

さまざまな

0.46

Activations Density 0.001%