INDEX
Explanations
discrepancies between summaries and their corresponding documents.
This neuron detects mentions of formal legal requests made by lawyers.
New Auto-Interp
Negative Logits
пут
-0.06
lup
-0.06
reck
-0.06
results
-0.06
trades
-0.06
pmat
-0.06
conti
-0.06
вп
-0.06
صرف
-0.06
nea
-0.06
POSITIVE LOGITS
(Block
0.07
бал
0.07
سال
0.07
iaomi
0.06
imate
0.06
_friends
0.06
dk
0.06
_Move
0.06
_down
0.06
medical
0.06
Activations Density 0.005%