INDEX

Explanations

pronoun-verb knowledge states

paste when ready

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

 söyl

0.46

 undertook

0.43

 здійсню

0.42

,_

0.41

 아멘

0.41

 BOARDSIZE

0.41

 يستخدم

0.41

 prayed

0.41

 groan

0.40

 oryg

0.40

POSITIVE LOGITS

 knowing

0.52

 know

0.51

 deem

0.48

 conoz

0.47

knowing

0.46

 deems

0.45

 knows

0.44

 daniel

0.44

 sabe

0.43

 tahu

0.43

Activations Density 0.142%