INDEX

Explanations

The neuron fires on contrastive or adversative discourse markers—words like “but,” “otherwise,” “however,” etc.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

0.41

ή

0.33

or

0.30

或其他

0.30

idespread

0.30

0.29

 hoặc

0.28

 eponymous

0.28

}}_{

0.28

POSITIVE LOGITS

jenigen

0.46

 ساتھ

0.43

 sooo

0.40

ूतक

0.38

YOU

0.38

 साथ

0.38

 যেই

0.38

сибо

0.38

 Natürlich

0.37

 soooo

0.37

Activations Density 0.604%