merging latest changes #1
Merged
Conversation
This file contains 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
* Add test, catch specific exception * fix rebase * fixed test after rebase * Modified test after rebase, addressed comments
…om disk. (#4321) * Fixed problem when saving and loading normalizer transformers that work on multidimensional vectors * Added test of normalizer loaded from disk for a multidimensional vector * Added test for backward compatibility with normalizer transformer with verWrittenCur: 0x00010001 * Added test reproducing the original scenario * Changed visibility of classes made for test
* Add missing linePragmas to ConsumeModel.tt Fixes #4400
* Integrate Image Classification API in AutoML. * Add code gen. * fix tests. * revert samples. * PR feedback. * Fix tests. * Fix tests. * add more tests. * Change default architecture to be resnet 50. * Change default architecture to be resnet 50. * Enable CV, Pass Validation set to Image Classfication Trainer, pipe image classification metrics callback to AutoML. * Add tensorflow binary dependency, remove data.model, and update tests. * cleanup. * minor fix of code gen * fix bug in image classfication state initialization. * Ignore bad images. * resolve merge conflicts. * PR feedback. * PR feedback. * PR feedback. * PR feedback. * PR feedback, fix tests.
Not disposing this cursor is causing the training file to remain locked even after training is completed. Fix #4361
Created Polynomial Learning Rate scheduler Created the polynomial learning rate scheduler based upon Tensorflow's implementation of polynomial lr scheduler. which can be found https://www.tensorflow.org/api_docs/python/tf/compat/v1/train/polynomial_decay
…lumns (#4403) * Updating the DatabaseLoaderTests to cover a single load column * Updating the DatabaseLoader to not force vector for single element columns
… between Microsoft.ML.TensorFlow and Microsoft.ML.Dnn (#4408) * Reverse dependency between Microsoft.ML.TensorFlow and Microsoft.ML.Dnn. * Reverse dependency between Microsoft.ML.TensorFlow and Microsoft.ML.Dnn. * Rename Microsoft.ML.Dnn nuget and namespace to Microsoft.ML.Vision. * rename VisionCatalog. * PR feedback. * PR feedback. * PR feedback. * cleanup. * PR feedback.
* Move Microsoft.ML.Vision to stable package. * Move Microsoft.ML.Vision to stable package.
…rk from FT (#4346) * extract TestFrameworkCommon project and remove reference to TestFramework project from Functional.Test project * update on TestFrameworkCommon * update on TestFrameworkCommon project * minor fix * fix reference * fix project reference * fix CI build error * sync latest code and resolve * take comments and made change * take comments and fix * modify sln file to build on azure pipeline
* Add CodeGen piece for Recommendation Task using MatrixFactorization
* Changed some defaults * Changed metrics callback default * metricsCallback will write to mlcontext log by default. The sample has been update to show how to get the output to console from the log. * deleted unnecessary comments * Addressed comments * Minor clean up. * Disable unstable test.
* Modified how data is saved to disk pre-trained meta files are now stored in one location always, this allows multiple runst to re-use the same meta file without having to redownload. Additionally added the ability to cleanup the temporary workspace used to train the model. This should prevent issues of running out of disk space when running multiple training session sequentially. * fixed extra line error. * changes based on comments. * comment fixes. * address comments. * fixed error * fix for when to download. * Remove special handling of inception as we have fixed the meta file to not have extras that need to be downloaded.
…on API. (#4430) * Do not pass validation set to Image Classification API. * remove metric callback since it is set to default in ImageClassification Trainer.
… free up unmanaged memory. (#4438) * Recreate workspace directory when fit() is called, fix documentation, free up unmanaged memory. * PR feedback.
…review. (#4414) * Bump release version to 1.5.0-preview and preview version to 0.17.0-preview. * disable app compat tool on Microsoft.ML.Vision.
* Upgraded OnnxRuntime to v1.0 and Google Protobuf to 3.10.1 * Fixed version number to 1.0.0 * Changed type checking for ReadOnlyMemory<char> and changed Char to char * Fixed string handling in CreateScalarNamedOnnxValue from code review comments * Disabled GPU codepath for running Onnx models until it is better supported in onnxruntime * Upgraded baseline files to match the new ORT version * Adjusted accuracy range for onnx test
* Aligned line ending to 85 char limit * Cleaned up Early Stopping sample * Added comments to ResnetV2101TransferLearningTrainTestSplit * Added comments to ResnetV2101TransferLearningEarlyStopping * Add comments to ImageClassificationDefault sample * Added comments to LearningRateSchedulingCifarResnetTransferLearning sample * added '.', formatted output. * fixed sample after rebase * fix ealy stopping sample * fix GetAbsolutePath function for samples. * remove unnecessary try catch block from samples. * review comments.
…#4453) * Increment stable API version and add new stable packages to the list. * Increment stable API version and add new stable packages to the list.
… Program.cs file (#4300) * Adding a Program.cs to ML.Samples.GPU project, different from the one in ML.Samples * Removing the content in the ML.Samples.GPU main method * Put again the original content of Program.cs in the ML.Samples.GPU project. * Remove original Program.cs from Samples.GPU references
* nightly build pipeline
* Stabilize the LR test Found issue with how we were using random for our ImageClassificationTrainer. This caused instability in our unit test, as we were not able to control the random seed. Modified the code to now use the same random object throughout, the trainer, thus allowing us to control the seed and therefor have predictable output.
*Changes in PredictionTransformer.cs and Calibrator.cs to fix the problem of the create methods not being called, to make CMP load its internal calibrator and predictor first so to assign the correct paramaters types and runtimes, and added a PredictionTransformerLoadTypeAttribute so that the binary prediction transformer knows what type to assign when loading a CMP as its internal model. *Added a working sample for using PFI with BPT and CMPB while loading a model from disk. This is based entirely in the original sample. *Added file CalibratedModelParametersTests.cs with tests that the CMPs modified in this PR are now being correctly loaded from disk. *Changed a couple of tests in LbfgsTests.cs that failed because they used casts that now return 'null'.
…actingTransformer (#4451) * Added onnx export support for string related transforms * Updated baseline test files A large portion of this commit is upgrading the baseline test files. The rest of the fixes deal with build breaks resulting from the upgrade of ORT version. * Fixed bugs in ValueToKeyMappingTransformer and added additional tests
…icrosoft.ML.ModelBuilder.AutoMLService.Gpu (#4474)
…Max (#4472) * Fixed model saving and loading of OneVersusAllTrainer to include SoftMax * Modified existing test to include SoftMax option * Modified test to verify both cases: when UseSoftmax is true and false
* Initial work for adding onnx export support for OptionalColumnTransform * Implemented support for optional initializers in OnnxTranformer to support OptionalColumnTransform * Fixed handling of double values and non-long numeric types * Removed redundant line * Updated review comment
* Fix a flaky Extensions.ML test. Make the reload model tests more resistant to timing changes. * PR feedback.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
We are excited to review your PR.
So we can do the best job, please check:
Fixes #nnnnin your description to cause GitHub to automatically close the issue(s) when your PR is merged.