INDEX

Explanations

directly or directly before

The neuron fires on words meaning “direct” (and its equivalents/translations)—e.g. “direct,” “directly,” Russian “собственно,” “справа,” Ukrainian “прямої,” 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

вли

-1.76

 oficialmente

-1.62

 accedi

-1.58

 viaj

-1.50

…»

-1.48

…」

-1.44

႐

-1.42

 somente

-1.41



-1.41

()");

-1.40

POSITIVE LOGITS

1.67

 maupun

1.66

↵

1.59

1.55

 first

1.55

What

1.54

1.53

1.52

 these

1.48

Activations Density 0.004%