INDEX

Explanations

code parsing with foreign language snippets

The neuron fires on programming/markup syntax tokens (keywords, punctuation, and symbols like “exports”, “function”, “@”, “meta”, “Label=”, “*/”, etc.), i.e. it detects code or tag structure rather than natural-language words.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 also

-1.24

-1.02

it

-0.96

bahnen

-0.95

 kapcs

-0.94

 και

-0.92

*{-

-0.92



-0.91

'}

-0.90

НЫЙ

-0.90

POSITIVE LOGITS

 еще

0.92

ඨ

0.90

 frontières

0.90

 всегда

0.87

 fidanz

0.86

 поэтому

0.85

的是

0.85

 imponer

0.85

הס

0.85

 primeras

0.84

Activations Density 0.001%