Merges for v0.6.4 #3
Closed
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
* Dockerfile for R container added * moved Dockerfile to new repo * use pip to install gluonts shell * Update examples/dockerfiles/Dockerfile.r Co-authored-by: Eduard Uffelmann <uffeledu@amazon.com> Co-authored-by: Jasper <schjaspe@amazon.de> Co-authored-by: Jasper <jasper.b.schulz@googlemail.com>
* added different methods to impute missing values * all the changes done: transformation improved, added as option in the constructor of all the models using it, added unit tests * added a class to organise the imputation of missing values, this is not a final version, but it is for review * addressed Jasper's changes, thanks * Changed the organisation to classes, made the change in the deepAR estimator. * reviewed and cleaned imputation methods for missing values * Update src/gluonts/model/deepvar/_estimator.py Co-authored-by: Lorenzo Stella <lorenzostella@gmail.com> * fixed the couple of issues and addressed comments * fixed a small typo * addressed the few comments of lorenzo, added descriptions in form of docstrings * hopefully corrected style check Co-authored-by: Francois-Xavier Aubet <aubetf@amazon.com> Co-authored-by: Lorenzo Stella <lorenzostella@gmail.com>
* Fix to_time_series_item in splitter. TimeSeriesItem takes List[List] not List[np.ndarray]. * Added test-case.
* Added mx module. Moved mxnet dependent modules into dedicated mx-module. Added stub files which import module-contents from new place. * Fix docs build. * Using import hack for backwards compatibility. * Ignoring deprecated shadow-modules in doctests.
* updated Pandas deprecation * updating Pandas deprecation * Simplify splitter.py convert two Timdelta objects into one Co-authored-by: Jasper <jasper.b.schulz@googlemail.com> Co-authored-by: Pouya Mohammadi <mpouya@amazon.com> Co-authored-by: Jasper <schjaspe@amazon.de> Co-authored-by: Jasper <jasper.b.schulz@googlemail.com>
* Adapted use of mx module. Applied isort. * fixup! Adapted use of mx module. Applied isort.
* Added README to shell. * Fixup.
* updated Pandas deprecation * updating Pandas deprecation * Simplify splitter.py convert two Timdelta objects into one Co-authored-by: Jasper <jasper.b.schulz@googlemail.com> * using right frequency for Timestamp Co-authored-by: Pouya Mohammadi <mpouya@amazon.com> Co-authored-by: Jasper <schjaspe@amazon.de> Co-authored-by: Jasper <jasper.b.schulz@googlemail.com>
* Removed doc requirements from setup requirements. * Install doc-requirements in jenkins.
* add shuffling function * add comments in the ShuffleIter class * ensure shuffle iter returns a iterator of the same size of base iter. * fix validation bug * change stop signal in ShuffleIter
* make distributions pickleable * fix issue with lds
…wslabs#890) * splitting features and uses DataEntry instead of TimeSeriesItem * fixed spacing in docstrings * fixed spacing in all the docstrings * updated splitter test cases to be compatible with type DataEntry * more concise splitter.py Co-authored-by: Jasper <jasper.b.schulz@googlemail.com> * More concise splitter.py Co-authored-by: Jasper <jasper.b.schulz@googlemail.com> * More descriptive splitter.py Co-authored-by: Jasper <jasper.b.schulz@googlemail.com> * reformatted for PR Co-authored-by: Pouya Mohammadi <mpouya@amazon.com> Co-authored-by: Jasper <schjaspe@amazon.de> Co-authored-by: Jasper <jasper.b.schulz@googlemail.com>
* added lagged transformation, rolling aggregation option and tests
…labs#900) * Corrected the formula for the stddev of the MixtureDistribution. * style correction * added test for the stddev * changed stddev test
* Removed TimeSeriesItem. * Removed Channels, save_dataset. * Removed unused imports. * Readded save_datasets but moved it into a method.
* add distribution_network * add distribution output * Update estimator.py * Update _estimator.py * Update estimator.py * Update _estimator.py * Update _estimator.py * Update _network.py * Update _estimator.py * Add files via upload * Add files via upload * Update _network.py * Add files via upload * Add files via upload * Add files via upload * Add files via upload * Add files via upload * Update test_model.py * Update test_model.py * Update test_model.py * Update test/model/simple_feedforward/test_model.py Co-authored-by: Lorenzo Stella <lorenzostella@gmail.com> * Update src/gluonts/model/simple_feedforward/_network.py Co-authored-by: Lorenzo Stella <lorenzostella@gmail.com> * Update _estimator.py * Update _estimator.py * Update test/model/simple_feedforward/test_model.py * Update test/model/simple_feedforward/test_model.py * Update test/model/simple_feedforward/test_model.py * Update test/model/simple_feedforward/test_model.py * Update test/model/simple_feedforward/test_model.py * Add files via upload * Add files via upload * Update test_model.py * Add files via upload * Add files via upload Co-authored-by: Lorenzo Stella <lorenzostella@gmail.com>
fix minor bugs where shuffle_buffer_length is taken as an implicit input Co-authored-by: Lorenzo Stella <lorenzostella@gmail.com>
For issue awslabs#906 The `theta` param. of Gamma gets scale by `scale *theta = alpha * mu * scale`. So only need to scale the `mu` .
* edge case with singleton values fix of imputation methods
* Added support for gzipped files. * Fixed gzip. * Added test for gzip.
* Adding support for past dynamic features, scaling and default parameters updates for MQCNN * Fixing black * Fixing train method signture for shuffle_buffer_length * Removed overloaded train method Co-authored-by: Danielle Robinson <dmmaddix@amazon.com>
* fix max_history of splitter and add tests * added rolling split assertions and tests
* Added StateFul to serde. * Address comment.
* include callback in trainer and example for warm starting * don't fetch extra sample batch
Co-authored-by: Jasper <schjaspe@amazon.de>
pytorch needs to be fetched from -f https://download.pytorch.org/whl/torch_stable.html
* Added ty.checked decorator. * Added Stateful and Stateless to `ty`. Disallow assignment to stateless objects outside of __init__.
* fixed max_window of time feature and added test case Co-authored-by: Lorenzo Stella <lorenzostella@gmail.com>
awslabs#1096) * Adding test_quantiles hyperparameter to the shell to specify the quantiles for evaluation * Using Quantile.parse method to handle 'p50' quantile case Co-authored-by: Danielle Robinson <dmmaddix@amazon.com>
* initial pytorch distribution output class Co-authored-by: Lorenzo Stella <lorenzostella@gmail.com>
* Refactored serde into a package. Removed `dump_json`, use pickle directly instead. Implemented encoding for `tuple` and `set`, they were encoded as lists before. * Renamed _code to _repr. * Removed print-debug. * Moved Stateful/Statelesl to serde. * Black. * Support for 3.6 * Added missing re-import. * Py36 support. * Removed example from docstring to pass test. * Make doctest happy.
Made `shell.sagemaker` more independent and moved gluon-ts specific things from it.
* Added nursery.glide.
* Updated pytest to v5. (awslabs#1102) * Updated pytest to v5. * Skip sagemaker-sdk tests on windows. * Ignore NaN values during training and throw a warning (training got stuck before) (awslabs#1104) checks for NaNs is now done after each batch and before updating the model parameters just skip the batch and continue with the epoch. More informative warning added. * cap pydantic version (awslabs#1115) * specialize plot method for QuantileForecast (awslabs#1114) * add item_id to forecast from seasonal naive (awslabs#1113) * Fix a few bugs about tensor shapes in default values for TFT implementation (awslabs#1093) * add more references (awslabs#1128) * added more references * corrected README * corrected REFERENCES Co-authored-by: Tim Januschowski <tjnsch@amazon.com> * Fixes awslabs#1106 (awslabs#1125) * Fixes awslabs#1106 * fixed TypedDict error * Skip tests which fail due to timeout. * Integrated comments. Co-authored-by: Lorenzo Stella <lorenzostella@gmail.com> * reduce number of batches used in test (awslabs#1131) * fix pandas usage and remove version cap (awslabs#1132) Co-authored-by: Jasper <schjaspe@amazon.de> Co-authored-by: Riccardo Grazzi <riccardo.grazz@gmail.com> Co-authored-by: Lorenzo Stella <lorenzostella@gmail.com> Co-authored-by: Xiaoyong Jin <x_jin@cs.ucsb.edu> Co-authored-by: tim <31938042+timoschowski@users.noreply.github.com> Co-authored-by: Tim Januschowski <tjnsch@amazon.com> Co-authored-by: kja <50238578+kaijennissen@users.noreply.github.com>
* Masking edge case fix (cherry picked from commit 1965d8e) Co-authored-by: Konstantinos Benidis <benidisk@gmail.com>
(cherry picked from commit 13dd13) Co-authored-by: Lorenzo Stella <lorenzostella@gmail.com>
* Added item_id to NPTS and Naive2 forecasts (awslabs#1173) * fixes in dataset mutability (awslabs#1171) Co-authored-by: kja <50238578+kaijennissen@users.noreply.github.com> Co-authored-by: Konstantinos Benidis <benidisk@gmail.com>
Schmedu
force-pushed
the
v0.6.x
branch
5 times, most recently
from
728da97
to
da74b19
Compare
* PyTorchPredictor serde (awslabs#1086) * Add equality operator for PytorchPredictor (awslabs#1190) * Add equality operator for PytorchPredictor * Make new black version happy Co-authored-by: Michael Bohlke-schneider <bohlkem@amazon.com> * fix pytorch predictor serde (awslabs#1194) Co-authored-by: Lorenzo Stella <lorenzostella@gmail.com> Co-authored-by: Michael Bohlke-Schneider <mschneid83@gmail.com> Co-authored-by: Michael Bohlke-schneider <bohlkem@amazon.com>
* Fix serde for np.dtype. (awslabs#1299) Co-authored-by: Jasper <schjaspe@amazon.de>
* Use broadcast_logical_or in inflated_beta (awslabs#1226) * Fix type error for using quantile_weights and add a proper Pytest (awslabs#1231) * Fixed MASE in N-BEATS: removed redundant factor (awslabs#1288) * Fixing bug where dropout was not used, also remove unused halt option (awslabs#1315) * Fixes for Python 3.8 (awslabs#1318) Co-authored-by: Dan Lucas <47125059+deejlucas@users.noreply.github.com> Co-authored-by: youngsuk0723 <pys1987@gmail.com> Co-authored-by: Danielle Robinson <dcmaddix@gmail.com> Co-authored-by: Riccardo Grazzi <riccardo.grazz@gmail.com> Co-authored-by: David Salinas <geoalgo@users.noreply.github.com>
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.
Merge commits: