INDEX

Explanations

Street name segments

This neuron fires on address and location tokens—e.g., street names, highway or route designators, state/region abbreviations, and other components of postal or geographic addresses.

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

and

-1.92

 protagonisti

-1.47

܊

-1.34

 மற்ற

-1.33

 bidra

-1.30

łko

-1.30

</th>

-1.30

龵

-1.29

 beni

-1.27

 miele

-1.27

POSITIVE LOGITS

BindingResult

1.89

</h2>

1.65

1.52

 stehenden

1.45

кос

1.41

 liegenden

1.41

注重

1.39

 Metern

1.38

﹟

1.38

高达

1.36

Activations Density 0.051%