Scale the negative binomial's gamma()

For issue awslabs#906 The `theta` param. of Gamma gets scale by `scale *theta = alpha * mu * scale`. So only need to scale the `mu` .
kashif · Jul 3, 2020 · cc09b02 · cc09b02
1 parent 5c2f4ce
commit cc09b02
Showing 1 changed file with 0 additions and 2 deletions.
diff --git a/src/gluonts/mx/distribution/neg_binomial.py b/src/gluonts/mx/distribution/neg_binomial.py
@@ -133,9 +133,7 @@ def distribution(
             return NegativeBinomial(mu, alpha)
         else:
             F = getF(mu)
-            scale = 1.0 + softplus(F, scale - 1.0)
             mu = F.broadcast_mul(mu, scale)
-            alpha = F.broadcast_add(alpha, F.broadcast_div(scale - 1, mu))
             return NegativeBinomial(mu, alpha, F)
 
     @property