Repository: incubator-systemml-website
Updated Branches:
refs/heads/master 0c170c572 -> b2f52ca65
[MINOR] Add security page under Apache
Closes #42.
Project: http://git-wip-us.apache.org/repos/asf/incubator-systemml-website/repo
Commit:
http://git-wip-us.apache.org/repos/asf/incubator-syste
Repository: incubator-systemml
Updated Branches:
refs/heads/master 6e7e8873a -> d127dfa2d
http://git-wip-us.apache.org/repos/asf/incubator-systemml/blob/d127dfa2/src/main/java/org/apache/sysml/runtime/matrix/data/LibMatrixDNN.java
[SYSTEMML-1411] Add CPU and GPU bias multiply operation
This operation is similar to bias_add, except that it performs
element-wise multiplication rather than addition. It avoids unnecessary
multiplication of ones and reshape. This pattern is common in deep
learning layers such as batch normalizat
Repository: incubator-systemml
Updated Branches:
refs/heads/master c663e31ce -> 6e7e8873a
[SYSTEMML-1403] Fixed the bug in GPU relu_maxpooling and also cleaned up
LibMatrixCUDA code.
Closes #432.
Project: http://git-wip-us.apache.org/repos/asf/incubator-systemml/repo
Commit:
http://git-wip
[SYSTEMML-1372] Extended code generator (celltmpl w/ matmult as root)
In order to handle the case where previous rewrites modified sum(X^2) -
which is fusable to any cellwise operations that created X - to
t(X)%*%X, we now also support dot products as roots of cellwise
templates (cellwise multiply
[SYSTEMML-1314] Robust memory handling for RDD parallelization
So far we only checked on RDD parallelization if the pinned matrix block
as well as the partitioned matrix fit together into the driver memory
budget; otherwise the matrix is exported (w/ a multi-threaded write in
binary block format
[MINOR] Robust hdfs file delete w/ retries (fix occasional test issues)
Project: http://git-wip-us.apache.org/repos/asf/incubator-systemml/repo
Commit:
http://git-wip-us.apache.org/repos/asf/incubator-systemml/commit/c663e31c
Tree: http://git-wip-us.apache.org/repos/asf/incubator-systemml/tree/c6
Repository: incubator-systemml
Updated Branches:
refs/heads/master 81090134d -> c663e31ce
[SYSTEMML-1364] Hardened code generator for sparse-unsafe outer products
Project: http://git-wip-us.apache.org/repos/asf/incubator-systemml/repo
Commit:
http://git-wip-us.apache.org/repos/asf/incubator-s
Repository: incubator-systemml
Updated Branches:
refs/heads/master f32bd8ebd -> 81090134d
[SYSTEMML-1370] Convert numpy to matrixblock by passing multiple blocks
Here is the code to test this functionality:
from systemml import MLContext, dml, convertToMatrixBlock
import pandas as pd
nr = 469