INDEX

Explanations

how and why questions

breaking down information

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 throughout

0.39

 Throughout

0.38

 Presidency

0.36

 Atau

0.34

 또는

0.33

 Northrop

0.33

</td>

0.33

ℓ

0.32

Or

0.32

POSITIVE LOGITS

how

0.89

 bagaimana

0.86

why

0.78

 كيفية

0.75

how

0.75

 hvordan

0.75

 কিভাবে

0.74

 cómo

0.73

如何

0.73

 কীভাবে

0.73

Activations Density 0.020%