INDEX

Explanations

weekend

The neuron fires on tokens that name transit‐service actions and timing (e.g. “overnight,” “weekend,” “provide,” “service”).

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

-1.52

tvguidetime

-1.39

 ANOTHER

-1.34

 ideen

-1.33

 heilig

-1.30

”、“

-1.27

期刊

-1.27

 Risiko

-1.27

疫情

-1.27

≥

-1.27

POSITIVE LOGITS

 here

1.47

 this

1.47

 these

1.45

in

1.36

on

1.33

 changing

1.33

 doing

1.29

 needing

1.23

 where

1.21

 with

1.19

Activations Density 0.034%