INDEX

Explanations

selecting descriptive terms and subsequent categories

loss of 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

 absolv

0.45

 مخلو

0.41

_{+}+

0.40

 ہتھیار

0.40

हरूको

0.38

 گرفتار

0.38

дел

0.38

 thử

0.37

 جهد

0.37

 monsieur

0.36

POSITIVE LOGITS

and

0.44

৩৯

0.43

bh

0.38

ф

0.36

And

0.36

 source

0.36

〝

0.35

ᱲ

0.35

秘

0.35

AVA

0.34

Activations Density 0.065%