INDEX

Explanations

specific words followed by punctuation

color, root, nausea, jasmine, distortion

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

 cuyos

0.43

ţ

0.42

Gł

0.39

 cuyas

0.39

̋

0.38

８

0.37

espół

0.36

 கண்க

0.36

 niekoľ

0.36

qualiter

0.36

POSITIVE LOGITS

 aspect

0.66

 requirement

0.63

 ব্যাপারটা

0.60

というのは

0.59

 mentioned

0.56

 stipulation

0.55

 কথাটা

0.55

 بودن

0.54

 issue

0.53

 বিষয়টা

0.52

Activations Density 0.272%