INDEX

Explanations

environment, atmosphere, context

The neuron fires on semantically “weighty” content words—key nouns and adjectives carrying factual or descriptive information in expository passages.

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

 podrán

0.94

 lediglich

0.90

 चाहेंगे

0.86

 possono

0.86

并未

0.86

自身の

0.86

未必

0.85

 हेतु

0.83

ለ

0.83

 underwhelming

0.81

POSITIVE LOGITS

 atmosphere

1.26

 ambiente

1.09

 atmos

1.07

 people

1.03

 ambience

1.00

 атмос

0.99

 suasana

0.96

 conviv

0.95

 людьми

0.95

 interferon

0.93

Activations Density 4.052%