INDEX

Explanations

checking for empty or existing

This neuron detects tokens that are part of code snippets (i.e. programming‐language keywords, identifiers, literals, 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

länder

-0.76

 provisions

-0.75

 Dickson

-0.75

 trains

-0.73

 opinión

-0.72

 Bereich

-0.71

jamas

-0.71

 instinct

-0.70

jsi

-0.70

 Çünkü

-0.70

POSITIVE LOGITS

any

1.23

isNotEmpty

1.18

Any

1.18

isEmpty

1.14

empty

1.11

NotEmpty

1.11

EMPTY

1.09

 empty

1.06

任何

1.06

 Empty

1.05

Activations Density 0.007%