INDEX

Explanations

producer names

The neuron strongly activates on all‐uppercase tokens and standalone numbers or punctuation (e.g. track headings, list numbers, and section markers).

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

 Gaga

-0.82

 slime

-0.79

 Giud

-0.77

ADA

-0.73

Candy

-0.72

 Minecraft

-0.71

Leh

-0.69

 Blockchain

-0.69

OTT

-0.69

ыр

-0.68

POSITIVE LOGITS

 scratched

0.80

 scratches

0.79

🟤

0.78

 jazz

0.77

 Bandcamp

0.77

Aes

0.75

funky

0.75

jaz

0.73

 sample

0.73

DriverManager

0.73

Activations Density 0.051%