INDEX

Explanations

a way to extend

This neuron fires on definitional or explanatory “as a” constructions (e.g. “described as a…,” “is nothing more than a…,” “run as a…”).

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

 mainstay

0.74

 worthwhile

0.69

 worthy

0.68

 favorite

0.66

 significant

0.65

 important

0.64

 seemingly

0.64

 beneficiary

0.64

 purportedly

0.62

 momentous

0.61

POSITIVE LOGITS

 extended

0.69

extended

0.67

 extension

0.66

 extensions

0.66

extensions

0.65

 extensión

0.64

延长

0.64

Extended

0.63

延長

0.63

 Extended

0.61

Activations Density 0.164%