INDEX

Explanations

words that will / remainder / remains / attacked

emphasizing specific details

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 stimulating

0.52

 stimulation

0.47

intended

0.46

পুলিশ

0.44

stimulating

0.44

 അവന്റെ

0.44

 haircut

0.43

允許

0.43

पुलिस

0.43

 stimulants

0.43

POSITIVE LOGITS

Ted

0.48

 Onda

0.47

vra

0.46

rl

0.45

 vele

0.45

 Stry

0.44

 Ashe

0.44

 Herr

0.44

 Gaye

0.44

絓

0.44

Activations Density 0.002%