INDEX
Explanations
closing braces in code snippets
New Auto-Interp
Negative Logits
and
-1.01
,
-0.90
-
-0.89
(
-0.87
in
-0.80
or
-0.79
(
-0.76
ins
-0.75
s
-0.75
one
-0.74
POSITIVE LOGITS
])))
2.09
.)}
2.08
}*/
2.05
]})
2.03
")}
2.03
})),
2.03
}}}}
2.02
")));
2.00
"]}
2.00
}))
1.99
Activations Density 1.172%