INDEX

Explanations

jpeg, progressive, abbreviations

The neuron is strongly activated by the lines and tokens that make up a C-style comment header (the “/* … /” block with leading “”-prefixed lines and metadata like Copyright, file descriptions, and directives).

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

 videos

-0.89

хии

-0.76

Videos

-0.72

 oprot

-0.72

 Videos

-0.70

 Flanders

-0.70

videos

-0.69

UBC

-0.69

isel

-0.68

цька

-0.67

POSITIVE LOGITS

Sos

0.88

Sos

0.85

MCU

0.79

jpeg

0.76

 progressive

0.76

dersfield

0.75

progressive

0.72

 Restart

0.72

Progressive

0.72

MCU

0.71

Activations Density 0.015%