INDEX

Explanations

specifying complex or nuanced situations

The neuron fires on content-carrying words (nouns, verbs, adjectives, etc.) and stays off for common function words (articles, prepositions, conjunctions).

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

 nouns

0.96

 irritability

0.96

 ideals

0.92

 refreshments

0.92

 leans

0.91

Jpn

0.88

 obscurity

0.86

 Idha

0.86

 binoculars

0.85

 militants

0.85

POSITIVE LOGITS

ad

0.88

لی

0.84

Blo

0.82

 capaces

0.76

ünft

0.76

⇉

0.73

ล

0.72

้าง

0.71

pca

0.71

ut

0.71

Activations Density 0.000%