INDEX

Explanations

definition or recommendation

This neuron responds to infrequent, content‐bearing words—especially uncommon nouns or proper nouns—i.e. low‐frequency terms.

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

 Even

0.83

 Much

0.83

And

0.82

But

0.80

 Though

0.80

 Like

0.79

 लेकिन

0.77

 Because

0.77

 aunque

0.77

 Although

0.75

POSITIVE LOGITS

ний

0.58

sthe

0.54

0.52

()]

0.50

жды

0.49

ferred

0.48

тата

0.48

iation

0.48

()];

0.47

_',

0.47

Activations Density 0.356%