Typed SDCA binary trainers #2506
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
wschin
force-pushed
the
typed-linear-trainers
branch
from
028d090 to
27536c3
Compare
wschin
force-pushed
the
typed-linear-trainers
branch
2 times, most recently
from
cc7766d to
b6135a9
Compare
Codecov Report
@@ Coverage Diff @@
## master #2506 +/- ##
==========================================
- Coverage 71.26% 71.24% -0.02%
==========================================
Files 798 798
Lines 141231 141236 +5
Branches 16112 16108 -4
==========================================
- Hits 100642 100627 -15
- Misses 36119 36143 +24
+ Partials 4470 4466 -4
|
Codecov Report
@@ Coverage Diff @@
## master #2506 +/- ##
=========================================
Coverage ? 71.24%
=========================================
Files ? 798
Lines ? 141236
Branches ? 16108
=========================================
Hits ? 100627
Misses ? 36143
Partials ? 4466
|
Ivanidzo4ka
reviewed
Feb 13, 2019
Ivanidzo4ka
reviewed
Feb 13, 2019
Ivanidzo4ka
reviewed
Feb 13, 2019
test/BaselineOutput/SingleDebug/SDCA/BinarySDCA-SmoothedHinge-CV-breast-cancer-out.txt
Outdated
Show resolved
Hide resolved
Ivanidzo4ka
reviewed
Feb 13, 2019
test/Microsoft.ML.Tests/Scenarios/Api/CookbookSamples/CookbookSamplesDynamicApi.cs
Outdated
Show resolved
Hide resolved
Ivanidzo4ka
reviewed
Feb 13, 2019
Ivanidzo4ka
reviewed
Feb 13, 2019
test/Microsoft.ML.Tests/TrainerEstimators/SymSgdClassificationTests.cs
Outdated
Show resolved
Hide resolved
Ivanidzo4ka
reviewed
Feb 13, 2019
Ivanidzo4ka
reviewed
Feb 13, 2019
test/BaselineOutput/SingleDebug/SDCALR/BinarySDCA-CV-breast-cancer-out.txt
Outdated
Show resolved
Hide resolved
Ivanidzo4ka
reviewed
Feb 13, 2019
sfilipi
reviewed
Feb 13, 2019
sfilipi
reviewed
Feb 13, 2019
sfilipi
reviewed
Feb 13, 2019
sfilipi
reviewed
Feb 13, 2019
sfilipi
reviewed
Feb 13, 2019
sfilipi
reviewed
Feb 13, 2019
sfilipi
reviewed
Feb 13, 2019
sfilipi
reviewed
Feb 13, 2019
sfilipi
reviewed
Feb 13, 2019
sfilipi
reviewed
Feb 13, 2019
sfilipi
reviewed
Feb 13, 2019
sfilipi
approved these changes
Feb 13, 2019
Fix some tests failed because of duplicated load names
1. Remove arguments to set up calibrator in all SDCA trainer 2. Clean up entry point base on argument's changes. 3. Update test files
TomFinley
reviewed
Feb 13, 2019
TomFinley
reviewed
Feb 13, 2019
wschin
force-pushed
the
typed-linear-trainers
branch
from
407bdde to
267c387
Compare
wschin
commented
Feb 14, 2019
docs/samples/Microsoft.ML.Samples/Dynamic/Trainers/BinaryClassification/SDCANonCalibrated.cs
Outdated
Show resolved
Hide resolved
Ivanidzo4ka
reviewed
Feb 14, 2019
docs/samples/Microsoft.ML.Samples/Dynamic/Trainers/BinaryClassification/SDCA.cs
Outdated
Show resolved
Hide resolved
Ivanidzo4ka
reviewed
Feb 14, 2019
Ivanidzo4ka
reviewed
Feb 14, 2019
Ivanidzo4ka
reviewed
Feb 14, 2019
TomFinley
reviewed
Feb 14, 2019
|
|
||
| namespace Microsoft.ML.Samples.Dynamic | ||
| { | ||
| public class SDCASupportVectorMachine |
There was a problem hiding this comment.
SDCASupportVectorMachine [](start = 17, length = 24)
This is a static class right? (Only if you happen to post another commit after this.)
There was a problem hiding this comment.
Hopefully #2548 can fix it.
|
Thank you @TomFinley. |
This was referenced Feb 14, 2019
ghost
locked as resolved and limited conversation to collaborators
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Make
SdcaBinaryTrainerstrongly-typed according to the type it produces. The existingSdcaBinaryTrainercan produce either calibrated linear model or linear model as mentioned in #2469. To fix #2469, we move common functionalities used in both cases toSdcaBinaryTrainerBase<T>whereTisCalibratedModelParametersBase<LinearBinaryModelParameters, PlattCalibratorfor calibrated case andLinearBinaryModelParametersotherwise. Top-level APIs are changed accordingly. For dynamic/static APIs, we have 4 SDCA binary trainers forIn addition, because we don't like auto-calibration, the output schema in uncalibrated cases should not contain a probability column. This PR also fixes this in the two derived trainer classes'
ComputeSdcaBinaryClassifierSchemaShape; the only case with a probability column generated is training logistic regression with Sdca. We also have two separatedSdcaCalibratedBinaryTrainer.OptionsandSdcaBinaryTrainer.Optionsderived fromBinaryArgumentBase. Legacy command-line tool and entry points are not affected by havingLegacySdcaBinaryTrainer.