INDEX

Explanations

journalists from

technical contexts

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

ếp

0.54

 Oven

0.51

 Largest

0.51

टी

0.50

 Lanes

0.48

زة

0.46

ات

0.46

ện

0.46

Tap

0.46

í

0.46

POSITIVE LOGITS

 cinematography

0.47

 serio

0.46

DetailUI

0.46

 ない

0.45

0.44

 ENGINEERING

0.44

<start_of_turn>

0.44

ColorLegend

0.44

<unused544>

0.43

Activations Density 0.001%