This repository has been archived by the owner on Jul 1, 2024. It is now read-only.
Keras-mxnet 2.2.4 release #193
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
…ras-team#10879) * Fix check equality between static shape and dynamic shape in test * Handle None in _keras_shape correctly
…as-team#10616) * Argument target_tensors from Model.compile: accept single tensor * Update target_tensors error message * Single tensor as target_tensors: Handle error for multi-output model. Added quick tests. * Added regex to tests
* Update TensorFlow as 1.9 on Travis * Test whether the time-out error comes from multiprocessing or not * Another try * Style fixes * Another try
* ENH: allow save_model/load_model to accept h5py.Group This is inspired by a bug report to h5py (h5py/h5py#1076) to allow the user to pass a Group or a File (which is a sub-class of Group) into save_model and load_model * TST: also test passing in an h5py.Group
…m#10921) * Raise warning if both class_weight and sample_weight given * Fix warning message style.
* GlobalAveragePooling1D supports masking * Use broadcast rather than repeat * Use static shape as far as possible
…onitored quantity (keras-team#10765) * Fix final weights from EarlyStopping class and add/modify test cases * Adds argument restore_best_weights to constructor, standardizes the docstring, changes the variable name from good_weights to best_weights and adds a message indicating that weights have been restored; adds tests for new argument
* Fix CI on TensorFlow * Another try * Another try * Another try
* Fix multi-output Siamese networks in multi_gpu_model * Remove Counter, add unit-test.
* Style fixes for enabling PEP8 501 * Minor fixes
* allow non-square weights to initialize with identity initializer * fix style * Fix docstring style.
* Keep the right val_steps * Add tests
…as-team#10888) * Made an abstract class for all the cropping layers. * Added the rank in the config of _Cropping. * Made self.cropping be a tuple of tuple in all cases (1d, 2d and 3d). * Removed the rank argument in _Cropping, allowing to remove the overwrides of get_config.
…ras-team#10952) * Update preprocessing/applications integration * Fix mobilenet_v2 filename. * Applications / preprocessing fixes. * Fix applications imports
…nputs and outputs. (keras-team#10939) * Added in the documentation an example of custom layer with multiple inputs and outputs. * Removed two unused imports and added assertions to make it easier to understand that methods arguments are lists.
* Working on improving tensor flow callbacks * Adding batch level TensorBoard logging (implementing the `on_batch_end` method to the TensorBoard class * Interim commit -- added notes. * Corrected stylistic issues -- brought to compliance w/ PEP8 * Added the missing argument in the test suite. * Added the possibility to choose how frequently tensorboard should log the metrics and losses. * Fixed the issue of the validation data not being displayed. * Fixed the issue about the callback not remembering when was the last time it wrote to the logs. * Removed the error check. * Used update_freq instead of write_step. * Forgot to change the constructor call.
* Graciously handle legacy Sequential configs * Remove warning
* Add `conv_transpose` into reference operations * Add a white space
### Summary This PR provides fix for `engine.training_generator.fit_generator` when `workers=0`. It used train data for validation and ignored `len(val_data)` when `val_data` was instance of `keras.Sequence` ### Related Issues [keras-team#10855](keras-team#10855) ### PR Overview - [x] This PR requires new unit tests [y/n] (make sure tests are included) - [ ] This PR requires to update the documentation [y/n] (make sure the docs are up-to-date) - [ ] This PR is backwards compatible [y/n] - [ ] This PR changes the current API [y/n] (all API changes need to be approved by fchollet)
* use mode arg * add loop test * add stricter condition * make dict behaviour similar
From line 1907 - 1918: when call tf.reshape(x, shape), (*before*)shape=(-1) will raise ValueError: Shape must be rank 1 but is rank 0 ... I think (-1) in python refers to the Rank-0 shape, though [-1] or (-1,) refers to the Rank-1 shape instead. It was also strange that it only raised error when 'image_data_format' was set to 'channels_first', not 'channels_last'.
* Improve addition_rnn example's code and comments * Fix trailing whitespace and newline * Update docstring
* Show time spent in the warning of slow progress * Wrap long line * Update warning message.
…keras-team#11226) Make fit_loop function easy to read and understand.
…perly (keras-team#11310) * fixing example format for multi_gpu_model * improving example 2 for multi_gpu_model * update example 3
|
Can you add link to the pickle issue that is opened with keras-team for mxnet backend? |
kalyc
reviewed
Oct 8, 2018
| 'pyyaml', | ||
| 'keras_applications==1.0.4', | ||
| 'keras_preprocessing==1.0.1'], | ||
| 'h5py', |
There was a problem hiding this comment.
is this compatible with older versions of h5py? we should probably keep the version requirement of h5py to prevent any breaking builds
There was a problem hiding this comment.
this change is from keras, let's keep it and it's not causing any build breaks: https://github.com/keras-team/keras/blob/master/setup.py
kalyc
approved these changes
Oct 8, 2018
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.
Summary
Related Issues
PR Overview