INDEX
Explanations
This neuron activates on descriptive adjectives modifying the following noun (e.g., “stone bridge,” “inland lake,” “Catholic school”).
New Auto-Interp
Negative Logits
�
-0.07
dao
-0.07
<f
-0.07
Snapshot
-0.07
multip
-0.07
проф
-0.07
woo
-0.07
methods
-0.06
İZ
-0.06
visto
-0.06
POSITIVE LOGITS
aquí
0.07
)];
0.06
) ↵ ↵ ↵
0.06
gratuits
0.06
Yoshi
0.06
))/(
0.06
_) ↵
0.06
]);
0.06
bli
0.06
);
0.06
Activations Density 0.055%