INDEX

Explanations

greek letters like lambda, kappa

The neuron fires on tokens that are (or contain) digits—that is, it identifies numbers and numeric sub‐tokens in the text.

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

 אלי

-0.88

 принадле

-0.79

%)$

-0.75

 engraçadas

-0.73

FFFFFF

-0.71

ماني

-0.71

Getenv

-0.70

karma

-0.69

бири

-0.68

dou

-0.67

POSITIVE LOGITS

1.05

0.91

0.90

 عدة

0.83

0.80

0.77

0.76

oczes

0.75

0.73

Activations Density 0.059%