INDEX

Explanations

visual loss rate

The neuron fires on technical parameter or property labels—i.e. words used as field names, option types, metric keys, component props, etc., in code or scientific descriptions.

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

[]:

-1.46



-1.41

\\

-1.41

\\

-1.36

 Años

-1.33

 }}{\

-1.33

Ǵ

-1.33

 fausse

-1.30

 nesse

-1.27

ᥬ

-1.27

POSITIVE LOGITS

 that

1.73

 regardless

1.54

 without

1.51

 until

1.50

 irrespective

1.40

or

1.38

we

1.32

 either

1.30

 especially

1.29

 uniquely

1.27

Activations Density 0.110%