INDEX

Explanations

left ventricle

The neuron strongly activates on standalone words that serve as headings or the first token of new sections, effectively marking paragraph or section titles.

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

 ово

0.59

妳

0.59

 стаби

0.59

湡

0.57

ᆞ

0.57

 your

0.56

'></

0.56

 />);

0.56

 আপনার

0.55

POSITIVE LOGITS

 Meanwhile

0.79

But

0.77

anwhile

0.77

But

0.76

Meanwhile

0.76

 Indeed

0.72

However

0.71

 그러나

0.71

 meantime

0.69

 However

0.68

Activations Density 0.036%