INDEX

Explanations

seeking advice or suggestions

This neuron fires on tokens involved in asking questions—especially meta-question phrases (e.g. “we … ask,” “?”) that solicit suggestions or further queries.

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

 same

-1.43

 каждого

-1.19

WHAT

-1.10

一辆

-1.09

available

-1.05

SOM

-1.05

Whereas

-1.05

same

-1.02

WHY

-1.02

 Acqu

-1.02

POSITIVE LOGITS

 this

1.83

 gewel

1.45

礎

1.43

ｫ

1.41

 that

1.41

or

1.41

 οποίο

1.37

 which

1.37

 или

1.30

驁

1.29

Activations Density 0.073%