GitHub user iyerr3 opened a pull request:
https://github.com/apache/madlib/pull/251
MLP: Simplify initialization of model coefficients
Changes:
1. Model initialization now happens in the C++ code instead of being
passed via Python.
2. Warm start coefficients are still passed from Python. These
coefficients are copied into the model. The vectorized form of the copy
raises an Eigen assertion when used with a MappedMatrix, possibly due
to the actual block size not being available via the Map.
3. The distinction between a "TaskState" and an "AlgoState" within a
State type has been removed in the MLPMiniBatchState. This demarcation
seemed to be confusing and didn't provide much abstraction.
Closes #251
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/iyerr3/incubator-madlib
bugfix/minibatch_init_debug
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/madlib/pull/251.patch
To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:
This closes #251
----
----
---
