INDEX

Explanations

Czech Republic, UK sizing, backup, cancel, Agent, Street

The neuron fires on structured-data tokens—uppercase labels, IDs, numeric values, and table-header or metadata terms.

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

</h2>

-1.45

ertia

-1.43

ကိုးက

-1.30

ಃ

-1.27

orul

-1.27

 garant

-1.26

 debo

-1.25

 eventi

-1.23

 débuts

-1.23

 finner

-1.22

POSITIVE LOGITS

↵↵↵

1.77

’

1.41

 Following

1.41

↵↵↵↵↵

1.30

 ikiwa

1.23

1.20

 commerciaux

1.20

 eenige

1.18

 років

1.17

ceuticals

1.13

Activations Density 0.333%