INDEX

Explanations

$$ and numbers

Option 1: Formal and Polished

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

帮你

0.41

ंसारी

0.40

ząt

0.39

orium

0.39

这才

0.39

clementine

0.38

зяржа

0.38

在美国

0.38

Открыть

0.37

mediation

0.37

POSITIVE LOGITS

$$\

0.61

$$

0.57

)$$

0.55

$$

0.53

 Idea

0.52

$$\

0.51

 cont

0.49

hw

0.48

 probl

0.46

0.45

Activations Density 0.005%