INDEX

Explanations

This neuron selectively activates on tokens that are numeric or part of capitalized names—that is, digits and proper-noun tokens.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ęła

0.75

onneur

0.74

(",")

0.70

 '.')

0.70

 தெரிவித்தார்

0.67

;,

0.66

τεύ

0.65

 '-')

0.63

 தெரிவி

0.63

ozygous

0.63

POSITIVE LOGITS

4.31

():

3.61

：

3.45

’:

3.39

!:

3.27

?:

3.22

':

3.17

*:

3.11

**:

3.11

+:

3.09

Activations Density 0.460%