INDEX

Explanations

code snippets and programming constructs

The neuron is essentially flagging contentful (non‐stop) words—i.e. nouns, verbs or adjectives carrying real “meaning” rather than common function words.

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

 aproximadamente

0.79

subsetneq

0.72

ponent

0.71

NN

0.71

serie

0.70

slightly

0.70

 approximately

0.70

approximately

0.70

 periode

0.70

ım

0.68

POSITIVE LOGITS

瞇

0.75

 wrestling

0.73

 म्हणा

0.70

 Mysteries

0.70

 दैट

0.70

Britt

0.69

Toc

0.69

лива

0.69

麥

0.68

rue

0.68

Activations Density 0.000%