INDEX

Explanations

code snippets and keywords

This neuron fires on formal legal‐document role labels (e.g. “Petitioner,” “Respondent,” “for”).

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

 получилось

-0.87

andidates

-0.84

 contribue

-0.81

brazilian

-0.80

 обеспечивает

-0.78

individuals

-0.77

 ويع

-0.75

somebody

-0.74

 rápidamente

-0.73

 douleur

-0.72

POSITIVE LOGITS

 autoestima

0.88

agra

0.86

手镯

0.82

 ملع

0.81

ğinde

0.81

mapp

0.79

toolkit

0.77

SYSTEM

0.76

norm

0.76

 приложения

0.75

Activations Density 0.120%