Adding getBestModel and getBestModelInfo to TuneHyperparametersModel #355

imatiach-msft · 2018-08-13T04:17:02Z

Adding getBestModel and getBestModelInfo to TuneHyperparametersModel (python API)

mmlspark-bot · 2018-08-13T19:10:20Z

FAIL Fail! — The build has failed. (a96fb410)

mmlspark-bot · 2018-08-13T19:33:37Z

FAIL Fail! — The build has failed. (a96fb410)

mmlspark-bot · 2018-08-13T20:14:56Z

FAIL Fail! — The build has failed. (a96fb410)

mmlspark-bot · 2018-08-13T21:24:08Z

FAIL Fail! — The build has failed. (a0f37312)

mmlspark-bot · 2018-08-13T22:31:35Z

PASS Pass! — The build has succeeded. (a0f37312)

MMLSpark 0.13.dev21+1.ga0f3731

This is a build for Github PR #355, changes:

[a0f3731] Ilya Matiach Adding getBestModel and getBestModelInfo to TuneHyperparameters

Source: imatiach-msft/mmlspark,
ilmat/tune-hyperparams-wrapper at revision a0f3731
(by Ilya Matiach [email protected]).
Build: MMLSpark, 1302375
(built by elbarzil-vm on elbarzil-vm, 2018-08-13 21:35)
Info:
0.13.dev21+1.ga0f3731: imatiach-msft/mmlspark/ilmat/tune-hyperparams-wrapper@a0f37312; MMLSpark#1302375
Queued by:
Ilya Matiach for Ilya Matiach
Maven package uploaded, use --packages com.microsoft.ml.spark:mmlspark_2.11:0.13.dev21+1.ga0f3731 and --repositories https://mmlspark.azureedge.net/maven.
PIP package uploaded.
R package uploaded.
HDInsight: Copy the link to this Script Action to setup this build on a cluster.
ARM Template: Use this ARM Template to setup a cluster with a gpu (in the portal).
- HDI Sub-Template: Use the HDI sub-template for just the hdi deployment (in the portal).
- GPU VM Sub-Template: Use the GPU sub-template for just the gpu vm deployment (in the portal).
- Convenient Deployment Script: Download this bash script or this powershell script, create a paramers file based on this template, and run as
```
./deploy-arm.sh ... -p <your-parameters>
```
Documentation uploaded.

mmlspark-bot · 2018-08-14T00:23:45Z

CANCEL Canceled! — The build has canceled. (7a04ead6)

mmlspark-bot · 2018-08-14T01:19:58Z

PASS Pass! — The build has succeeded. (7a04ead6)

MMLSpark 0.13.dev21+1.g7a04ead

This is a build for Github PR #355, changes:

[7a04ead] Ilya Matiach Adding getBestModel and getBestModelInfo to TuneHyperparameters and fixed autogenerated code for model name

Source: imatiach-msft/mmlspark,
ilmat/tune-hyperparams-wrapper at revision 7a04ead
(by Ilya Matiach [email protected]).
Build: MMLSpark, 1303390
(built by elbarzil-vm on elbarzil-vm, 2018-08-14 00:24)
Info:
0.13.dev21+1.g7a04ead: imatiach-msft/mmlspark/ilmat/tune-hyperparams-wrapper@7a04ead6; MMLSpark#1303390
Queued by:
Ilya Matiach for Ilya Matiach
Maven package uploaded, use --packages com.microsoft.ml.spark:mmlspark_2.11:0.13.dev21+1.g7a04ead and --repositories https://mmlspark.azureedge.net/maven.
PIP package uploaded.
R package uploaded.
HDInsight: Copy the link to this Script Action to setup this build on a cluster.
ARM Template: Use this ARM Template to setup a cluster with a gpu (in the portal).
- HDI Sub-Template: Use the HDI sub-template for just the hdi deployment (in the portal).
- GPU VM Sub-Template: Use the GPU sub-template for just the gpu vm deployment (in the portal).
- Convenient Deployment Script: Download this bash script or this powershell script, create a paramers file based on this template, and run as
```
./deploy-arm.sh ... -p <your-parameters>
```
Documentation uploaded.

mhamilton723 · 2018-08-15T20:23:28Z

src/codegen/src/main/scala/WrapperGenerator.scala

            }
            val (modelClass, modelQualifiedClass) = sc.getSuperclass.getSimpleName match {
              case "Estimator" => getModelFromGenericType(typeArgs.head)
+              case "ProbabilisticClassifier" => getModelFromGenericType(typeArgs(2))


Why do we need to specifically code against this API here? seems like this muddies the abstraction and is not needed becase a Probabbalistic Classifier is a predictor

I needed probabilistic classifier to get the model type, otherwise in the autogenerated lightgbm wrappers the model was "M" instead of lightgbmclassificationmodel

mhamilton723 · 2018-08-15T20:23:32Z

src/codegen/src/main/scala/WrapperGenerator.scala

@@ -23,6 +23,8 @@ abstract class WrapperGenerator {

  def wrapperName(myClass: Class[_]): String

+  def modelWrapperName(myClass: Class[_], modelName: String): String
+


does this need to also take in the modelName?

the model wrapper needs to take in the model name because that is what it returns, it uses the class to determine whether to put an underscore at the beginning

mhamilton723 · 2018-08-15T20:25:38Z

src/lightgbm/src/main/python/LightGBMRegressor.py

@@ -9,7 +9,7 @@
    basestring = str

 from mmlspark._LightGBMRegressor import _LightGBMRegressor
-from mmlspark._LightGBMRegressor import M
+from mmlspark._LightGBMRegressor import _LightGBMRegressionModel


This is nice

…ixed autogenerated code for model name

mmlspark-bot · 2018-08-18T00:30:26Z

FAIL Fail! — The build has failed. (c1ea31d1)

mmlspark-bot · 2018-08-18T03:14:12Z

PASS Pass! — The build has succeeded. (c1ea31d1)

MMLSpark 0.13.dev23+1.gc1ea31d

This is a build for Github PR #355, changes:

[c1ea31d] Ilya Matiach Adding getBestModel and getBestModelInfo to TuneHyperparameters and fixed autogenerated code for model name

Source: imatiach-msft/mmlspark,
ilmat/tune-hyperparams-wrapper at revision c1ea31d
(by Ilya Matiach [email protected]).
Build: MMLSpark, 1328977
(built by elbarzil-vm on elbarzil-vm, 2018-08-18 02:13)
Info:
0.13.dev23+1.gc1ea31d: imatiach-msft/mmlspark/ilmat/tune-hyperparams-wrapper@c1ea31d1; MMLSpark#1328977
Queued by:
Ilya Matiach for Ilya Matiach
Maven package uploaded, use --packages com.microsoft.ml.spark:mmlspark_2.11:0.13.dev23+1.gc1ea31d and --repositories https://mmlspark.azureedge.net/maven.
PIP package uploaded.
R package uploaded.
HDInsight: Copy the link to this Script Action to setup this build on a cluster.
ARM Template: Use this ARM Template to setup a cluster with a gpu (in the portal).
- HDI Sub-Template: Use the HDI sub-template for just the hdi deployment (in the portal).
- GPU VM Sub-Template: Use the GPU sub-template for just the gpu vm deployment (in the portal).
- Convenient Deployment Script: Download this bash script or this powershell script, create a paramers file based on this template, and run as
```
./deploy-arm.sh ... -p <your-parameters>
```
Documentation uploaded.

imatiach-msft force-pushed the ilmat/tune-hyperparams-wrapper branch from a96fb41 to a0f3731 Compare August 13, 2018 20:24

imatiach-msft force-pushed the ilmat/tune-hyperparams-wrapper branch 2 times, most recently from b68dc3b to 7a04ead Compare August 14, 2018 00:23

mhamilton723 requested changes Aug 15, 2018

View reviewed changes

mhamilton723 previously approved these changes Aug 17, 2018

View reviewed changes

Adding getBestModel and getBestModelInfo to TuneHyperparameters and f…

c1ea31d

…ixed autogenerated code for model name

imatiach-msft dismissed mhamilton723’s stale review via c1ea31d August 17, 2018 23:32

imatiach-msft force-pushed the ilmat/tune-hyperparams-wrapper branch from 7a04ead to c1ea31d Compare August 17, 2018 23:32

mhamilton723 approved these changes Aug 18, 2018

View reviewed changes

mhamilton723 merged commit d287118 into microsoft:master Aug 18, 2018

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Adding getBestModel and getBestModelInfo to TuneHyperparametersModel #355

Adding getBestModel and getBestModelInfo to TuneHyperparametersModel #355

imatiach-msft commented Aug 13, 2018

mmlspark-bot commented Aug 13, 2018

mmlspark-bot commented Aug 13, 2018

mmlspark-bot commented Aug 13, 2018

mmlspark-bot commented Aug 13, 2018

mmlspark-bot commented Aug 13, 2018

mmlspark-bot commented Aug 14, 2018

mmlspark-bot commented Aug 14, 2018

mhamilton723 Aug 15, 2018

imatiach-msft Aug 16, 2018

mhamilton723 Aug 15, 2018

imatiach-msft Aug 16, 2018

mhamilton723 Aug 15, 2018

imatiach-msft Aug 16, 2018

mmlspark-bot commented Aug 18, 2018

mmlspark-bot commented Aug 18, 2018

		@@ -23,6 +23,8 @@ abstract class WrapperGenerator {

		def wrapperName(myClass: Class[_]): String

		def modelWrapperName(myClass: Class[_], modelName: String): String

Adding getBestModel and getBestModelInfo to TuneHyperparametersModel #355

Adding getBestModel and getBestModelInfo to TuneHyperparametersModel #355

Conversation

imatiach-msft commented Aug 13, 2018

mmlspark-bot commented Aug 13, 2018

mmlspark-bot commented Aug 13, 2018

mmlspark-bot commented Aug 13, 2018

mmlspark-bot commented Aug 13, 2018

mmlspark-bot commented Aug 13, 2018

MMLSpark 0.13.dev21+1.ga0f3731

This is a build for Github PR #355, changes:

mmlspark-bot commented Aug 14, 2018

mmlspark-bot commented Aug 14, 2018

MMLSpark 0.13.dev21+1.g7a04ead

This is a build for Github PR #355, changes:

mhamilton723 Aug 15, 2018

Choose a reason for hiding this comment

imatiach-msft Aug 16, 2018

Choose a reason for hiding this comment

mhamilton723 Aug 15, 2018

Choose a reason for hiding this comment

imatiach-msft Aug 16, 2018

Choose a reason for hiding this comment

mhamilton723 Aug 15, 2018

Choose a reason for hiding this comment

imatiach-msft Aug 16, 2018

Choose a reason for hiding this comment

mmlspark-bot commented Aug 18, 2018

mmlspark-bot commented Aug 18, 2018

MMLSpark 0.13.dev23+1.gc1ea31d

This is a build for Github PR #355, changes: