INDEX

Explanations

variable, dates, or steps

The main thing this neuron does is detect standalone numeric tokens (i.e. sequences of digits).

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 ένας

0.65

 který

0.64

 belladone

0.63

 whose

0.60

 Moż

0.59

 ktorý

0.58

 protégé

0.57

 maître

0.57

 conçu

0.57

 الذي

0.56

POSITIVE LOGITS

ﺭ

0.55

зать

0.54

',

0.53

ков

0.51

тинен

0.51

ྪ

0.50

'=>

0.50



0.50

()

0.50

"')

0.50

Activations Density 0.029%