INDEX

Explanations

Orientational, future material, burglars

The neuron fires on snippets of PHP source code—i.e. programming‐language tokens and syntax (namespaces, backslashes, arrows, semicolons, etc.).

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

 incendios

-0.80

 različ

-0.77

INCIP

-0.75

 slick

-0.75

 zipped

-0.75

новь

-0.74

Detected

-0.74

warnai

-0.72

babwe

-0.72

 Erişim

-0.72

POSITIVE LOGITS

러

0.85

 Safe

0.79

safe

0.78

Safe

0.77

filter

0.77

ഊ

0.76

过滤

0.74

 safe

0.74

[['

0.73

ANN

0.71

Activations Density 0.008%