INDEX

Explanations

numbers and counts

The neuron fires on numeric tokens, i.e. it detects numbers (like dates, times, counts, or other digit sequences).

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 convolution

0.64

0.60

 بعنوان

0.57

 индек

0.56

bla

0.55

 hiper

0.55

 první

0.54

 convolutions

0.54

∫

0.54

0.53

POSITIVE LOGITS

対処

0.61

JU

0.61

۲۰

0.59

OJ

0.55

fz

0.54

satunya

0.53

四十

0.53

मात्र

0.53

 lẻ

0.53

㉖

0.52

Activations Density 0.148%