mx.nd.multi.lars
¶
Description¶
Compute the LARS coefficients of multiple weights and grads from their sums of square”
Arguments¶
Argument |
Description |
---|---|
|
NDArray-or-Symbol. Learning rates to scale by LARS coefficient |
|
NDArray-or-Symbol sum of square of weights arrays |
|
NDArray-or-Symbol sum of square of gradients arrays |
|
NDArray-or-Symbol weight decays |
|
float, required. LARS eta |
|
float, required. LARS eps |
|
float, optional, default=1. Gradient rescaling factor |
Value¶
out
The result mx.ndarray
Link to Source Code: http://github.com/apache/incubator-mxnet/blob/1.6.0/src/operator/contrib/multi_lars.cc#L37