INDEX

Explanations

numbers and specific terms

The neuron fires on content‐bearing tokens—especially domain-specific technical terms, names of measured quantities, identifiers, and numbers—distinguishing them from common function words.

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

likation

-1.14

 コットン

-1.09

 NATURE

-1.06

umque

-1.06

 HUMAN

-1.05

糁

-1.05

){$\

-1.05

 Reſ

-1.04

施展

-1.04

</tbody>

-1.03

POSITIVE LOGITS

 after

1.67

on

1.45

 eight

1.43

 four

1.40

in

1.32

 nine

1.29

had

1.27

 Using

1.27

two

1.26

 three

1.22

Activations Density 0.001%