INDEX
Explanations
descriptions of food flavors and textures
New Auto-Interp
Negative Logits
Loving
-0.16
Bench
-0.15
conv
-0.15
iyim
-0.14
394
-0.14
loving
-0.14
ooter
-0.14
slower
-0.14
øy
-0.14
elter
-0.14
POSITIVE LOGITS
wat
0.29
Wat
0.27
wat
0.25
rubber
0.23
metallic
0.22
bitterness
0.21
Wat
0.21
gritty
0.21
bitter
0.21
gre
0.20
Activations Density 0.075%