INDEX

Explanations

service or services

The neuron strongly activates on the word “service” (and its inflected forms like “services,” “serviced,” etc.).

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

 distribuidor

-1.51

 suplemento

-1.45

鴟

-1.41

 enhances

-1.40

所谓的

-1.38

 âme

-1.37

 محصولات

-1.33

 bibit

-1.33

 joga

-1.33

 corredor

-1.31

POSITIVE LOGITS

and

1.65

 هذا

1.52

 Instead

1.51

It

1.46

さつ

1.45

BUT

1.45

And

1.43

なりません

1.42

 Then

1.41

 Immediately

1.38

Activations Density 0.028%