INDEX

Explanations

often

The neuron strongly activates on content words (usually nouns or noun phrases) that begin a new sentence or section—i.e. capitalized words at sentence-start.

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

 ہے

0.71

是一

0.65

AUTH

0.62

remarkable

0.61

lighting

0.61

 है

0.59

嬉しい

0.59

ляет

0.59

是一个

0.59

לה

0.59

POSITIVE LOGITS

 lahko

0.70

 biasanya

0.67

 обычно

0.66

often

0.64

 ondas

0.64

 spesso

0.63

 সাধারণত

0.63

往往

0.63

 பொதுவாக

0.62

 COLLISIONS

0.62

Activations Density 0.055%