INDEX
Explanations
Reviews or technical language
The neuron fires on multi-syllable, content-heavy terms (especially technical or scientific jargon and gerunds) rather than on common function words.
New Auto-Interp
Negative Logits
savaş
-0.06
Convert
-0.06
runners
-0.06
losing
-0.06
MAGIC
-0.06
notations
-0.06
antity
-0.06
ها
-0.06
ervas
-0.06
amples
-0.06
POSITIVE LOGITS
��
0.07
yogurt
0.07
От
0.07
wsp
0.06
BAD
0.06
악
0.06
其
0.06
Без
0.06
آبی
0.06
_REL
0.06
Activations Density 0.479%