INDEX

Explanations

intend to, last time, both loops

aiming for transparency

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

の結果

0.54

 ಶಿ

0.48

窰

0.47

 curving

0.45

வேண்டும்

0.45

 Klima

0.45

部分的

0.44

 Söz

0.44

オシャレ

0.44

 Natasha

0.44

POSITIVE LOGITS

হিত

0.47

 brat

0.46

staw

0.45

লেই

0.43

Transparency

0.41

brat

0.41

ใส

0.41

saw

0.40

ipfs

0.40

splice

0.39

Activations Density 0.000%