INDEX
Explanations
Informal conversational writing
The neuron selectively fires on tokens found in the reader‐submitted comment sections (i.e. it detects conversational comment text rather than the main post).
New Auto-Interp
Negative Logits
Twilight
-0.07
Each
-0.07
нед
-0.06
pony
-0.06
enstein
-0.06
_coupon
-0.06
.direction
-0.06
>j
-0.06
(req
-0.06
exels
-0.06
POSITIVE LOGITS
KeyboardInterrupt
0.07
аліст
0.06
_entropy
0.06
inesis
0.06
ической
0.06
�
0.06
đúng
0.06
reich
0.06
.dtp
0.06
noteq
0.06
Activations Density 0.054%