INDEX
Explanations
awards,SAG,World War,Games,Gemma
Tokens that mark dates, numbers, or named news/events (e.g., event titles, competitions, conferences, and other time- or event-related proper nouns).
New Auto-Interp
Negative Logits
uses
0.38
Uses
0.35
using
0.35
Uses
0.34
switching
0.33
utilisant
0.33
overuse
0.33
serialized
0.32
sử
0.31
사용하는
0.31
POSITIVE LOGITS
vej
0.35
テール
0.31
kemarin
0.30
कमेंट
0.29
seh
0.28
wonderland
0.28
달라
0.28
edizione
0.28
astal
0.28
mendatang
0.28
Activations Density 0.108%