keras-team / keras Public

1 Release published by 1 person

• v2.7.0-rc2 Keras Release 2.7.0 RC2

  published Oct 28, 2021

3 Pull requests merged by 3 people

• Add efficientnet v2 to keras.applications

  #14935 merged Oct 29, 2021

• Disable the failing tests due to numpy 1.20 change

  #15552 merged Oct 27, 2021

• Added layer trainable information in model summary

  #15459 merged Oct 24, 2021

5 Pull requests opened by 4 people

• Fix typos

  #15543 opened Oct 24, 2021

• Google only changes

  #15550 opened Oct 25, 2021

• chore: replace os.path.join with tf.io.gfile.join

  #15551 opened Oct 26, 2021

• Add check for ground truth label range in function `binary_crossentropy()`

  #15554 opened Oct 27, 2021

• Keras IoU Metrics.

  #15567 opened Oct 29, 2021

23 Issues closed by 9 people

• Invalid Argument Error

  #15518 closed Oct 29, 2021

• Setting None to fname parameter in Keras utils get_file() throws an error

  #15498 closed Oct 29, 2021

• Unable to use Keras data generator with Functional API model

  #14916 closed Oct 29, 2021

• Unspecified data type in image_dataset_from_directory

  #15500 closed Oct 28, 2021

• image_dataset_from_directory label list appear to be assigned in a random order

  #14416 closed Oct 28, 2021

• AttributeError: 'tensorflow.python.framework.ops.EagerTensor' object has no attribute 'assign'

  #14175 closed Oct 28, 2021

• ValueError: A `Concatenate` layer requires inputs with matching shapes except for the concat axis. Got inputs shapes: [(None, 12, 12, 128), (None, 12, 12, 32), (None, 12, 12, 32), (None, 12, 12, 64)]

  #14346 closed Oct 28, 2021

• 3D Autoencoder learnt only 1 dimension well

  #15558 closed Oct 28, 2021

• OSError: Unable to open file (unable to open file: name = 'best_model.hdf5', errno = 2, error message = 'No such file or directory', flags = 0, o_flags = 0)

  #14074 closed Oct 27, 2021

• Negative dimension size caused by subtracting 2 from 1 for 'pool2/AvgPool' (op: 'AvgPool') with input shapes: [?,1,56,128].

  #14095 closed Oct 27, 2021

• Can no longer use file buffer to save/load model

  #14411 closed Oct 27, 2021

• CNN regression model gives same output for all inputs

  #14458 closed Oct 27, 2021

• GPU Error: Check failed: work_element_count > 0 (0 vs. 0)

  #15296 closed Oct 27, 2021

• keras accuracy goes down dramatically after loading saved model.

  #14231 closed Oct 27, 2021

• Transfer learning model only predict same class, even though has 96-99% val_acc

  #14185 closed Oct 27, 2021

• fit_generator issue

  #14177 closed Oct 26, 2021

• Generator validation loss is significantly different from the sum of all the validation losses

  #14044 closed Oct 26, 2021

• Implementing a version of the (Sparse) Categorical Cross Entropy Loss that makes use of Integer-Indexed predictions.

  #14506 closed Oct 26, 2021

• Keras custom layer issue

  #15470 closed Oct 26, 2021

• Input 0 of layer denseC1 is incompatible with the layer: : expected min_ndim=2, found ndim=1.

  #15521 closed Oct 24, 2021

• Unable to resume training when using steps per epoch

  #15538 closed Oct 24, 2021

• How to add tf.keras.layers.AdditiveAttention in my model?

  #14246 closed Oct 23, 2021

• How to put data preprocessing layer into model

  #14261 closed Oct 23, 2021

10 Issues opened by 9 people

• TypeError: 'KerasTensor' object cannot be interpreted as an integer

  #15566 opened Oct 29, 2021

• TensorFlow Keras SavedModel with custom layers throws a TypeError after being saved and loaded twice

  #15565 opened Oct 29, 2021

• MultiHeadAttention layer save and load changes weight

  #15560 opened Oct 28, 2021

• Support building without Bazel

  #15559 opened Oct 28, 2021

• Cannot load back model with no-op Concatenate layer

  #15547 opened Oct 25, 2021

• Simple SSD, face detector, training ok with GradientTape BUT loss doesn't decrease if I train an identical model with Model.fit()

  #15546 opened Oct 25, 2021

• Add Swin-Transformer to keras.applications

  #15545 opened Oct 25, 2021

• keras get_custom_objects()

  #15544 opened Oct 25, 2021

• AttributeError: 'Functional' object has no attribute '_layers'

  #15542 opened Oct 24, 2021

• Periodic delays of TF inference time in batch process

  #15541 opened Oct 24, 2021

64 Unresolved conversations

Sometimes conversations happen on old items that aren’t yet closed. Here is a list of all the Issues and Pull Requests with unresolved conversations.

• InvalidArgumentError: slice index 22 of dimension 0 out of bounds. (reproducible code is given / colab).

  #15522 commented on Oct 28, 2021 • 4 new comments

• Update the contribution guide to include a applications section

  #15447 commented on Oct 28, 2021 • 4 new comments

• Recieve list of all outputs as input to a custom loss function.

  #14140 commented on Oct 26, 2021 • 2 new comments

• [Feature Request] ConvND and ConvNDTranspose layers

  #15490 commented on Oct 28, 2021 • 2 new comments

• A model with TextVectorization layer can not be uploaded

  #15443 commented on Oct 28, 2021 • 2 new comments

• Adding RegNets to tf.keras.applications

  #15419 commented on Oct 29, 2021 • 2 new comments

• confusion in setting steps_per_epoch and validation_steps for model.fit_generator

  #14595 commented on Oct 23, 2021 • 1 new comment

• How to load weights from .h5 file for WGAN-GP? I follow the official tutorial but don't know how to load weights for each epoch.

  #14631 commented on Oct 23, 2021 • 1 new comment

• Masking layer slows down fitting the model

  #14452 commented on Oct 23, 2021 • 1 new comment

• Keras training problem - no overfit with a small number of data

  #14606 commented on Oct 23, 2021 • 1 new comment

• Convert the six lower triangular values of the matrix into a diagonal matrix in keras

  #14674 commented on Oct 23, 2021 • 1 new comment

• Intermediate activations extractor for nested sequential

  #14245 commented on Oct 24, 2021 • 1 new comment

• tf.data gives Cannot batch tensors with different shapes in component 0

  #14532 commented on Oct 24, 2021 • 1 new comment

• How to use result of LSTM after fitting the model

  #14381 commented on Oct 24, 2021 • 1 new comment

• Stuck in `fit` function when I use 3080 with Cuda 11.0 and TensorFlow 1

  #14441 commented on Oct 24, 2021 • 1 new comment

• Need to save and reuse a data generator

  #14276 commented on Oct 25, 2021 • 1 new comment

• Triplet Loss Slice Index Error

  #14263 commented on Oct 25, 2021 • 1 new comment

• keras InvalidArgumentError: Incompatible shapes: [10,1] vs. [10,60,60,3600]

  #14268 commented on Oct 25, 2021 • 1 new comment

• Keras custom data generator giving dimension errors with multi input and multi output( functional api model)

  #14274 commented on Oct 25, 2021 • 1 new comment

• cannot import name 'detect_lp'

  #14208 commented on Oct 25, 2021 • 1 new comment

• AssertionError in keras:

  #14204 commented on Oct 25, 2021 • 1 new comment

• CNN visualization with keras and a pretrained model (resnet50)

  #14182 commented on Oct 25, 2021 • 1 new comment

• ValueError: An operation has `None` for gradient. Please make sure that all of your ops have a gradient defined (i.e. are differentiable)

  #14171 commented on Oct 25, 2021 • 1 new comment

• ValueError: Could not interpret optimizer identifier: <rl.util.AdditionalUpdatesOptimizer object at 0x0000026DBDBD5748>

  #14646 commented on Oct 26, 2021 • 1 new comment

• ValueError: generator yielded an element of shape () where an element of shape (None, None) was expected.

  #14308 commented on Oct 26, 2021 • 1 new comment

• Seq2seq with Attention and TeacherForcing

  #15456 commented on Oct 26, 2021 • 1 new comment

• tensoflow.keras.preprocessing.image_dataset_from_directory doesn't recognize there are files in a directory when labels are supplied as a list/tupple comp:keras stat:awaiting

  #15501 commented on Oct 26, 2021 • 1 new comment

• How to verify if masking is appropriately applied? LSTM performance drop with short sequences.

  #14242 commented on Oct 26, 2021 • 1 new comment

• Regression in model.optimizer.get_gradients() with TF 2.2

  #14066 commented on Oct 26, 2021 • 1 new comment

• I am having trouble save the Keras model.

  #14168 commented on Oct 26, 2021 • 1 new comment

• Keras Mask Issue When Loading Model With Custom Layer

  #15510 commented on Oct 26, 2021 • 1 new comment

• Load_weights problem. NotFoundError: Key optimizer/iter/.ATTRIBUTES/VARIABLE_VALUE not found in checkpoint [Op:RestoreV2]

  #14401 commented on Oct 26, 2021 • 1 new comment

• WGAn

  #14327 commented on Oct 27, 2021 • 1 new comment

• KeyError: 'input_4'

  #14322 commented on Oct 27, 2021 • 1 new comment

• Keras saved model returns different result than original model using Batch Normalization on multiple GPUs (Distributed training)

  #14372 commented on Oct 27, 2021 • 1 new comment

• Combining multiple models with multiple outputs.

  #14279 commented on Oct 27, 2021 • 1 new comment

• 30:20WARNING:tensorflow:Early stopping conditioned on metric `val_accuracy` which is not available. Available metrics are: WARNING:tensorflow:Reduce LR on plateau conditioned on metric `val_accuracy` which is not available. Available metrics are: lr

  #14183 commented on Oct 27, 2021 • 1 new comment

• how to create custom loss function which can slice model input data in keras

  #14688 commented on Oct 27, 2021 • 1 new comment

• Error randomly appears while training using a Custom Data Generator and specifying workers

  #14722 commented on Oct 27, 2021 • 1 new comment

• KERAS stops executing randomly while adding first layer inside docker container

  #14811 commented on Oct 27, 2021 • 1 new comment

• Using the same seed kwarg returns different values between GlorotUniform and GlorotUniformV2

  #15475 commented on Oct 27, 2021 • 1 new comment

• deadlock in model.predict_generator() with use_multiprocessing=True

  #14332 commented on Oct 27, 2021 • 1 new comment

• Huber loss crashes with exit code 0xC0000005 in Windows

  #14337 commented on Oct 27, 2021 • 1 new comment

• Predict on a live video feed for pretrained mobilenet model

  #14347 commented on Oct 27, 2021 • 1 new comment

• Deeplearning4J: InvalidKerasConfigurationException: Expected model class name Model (found Functional).

  #14360 commented on Oct 27, 2021 • 1 new comment

• consistency of keras and tensorflow

  #14385 commented on Oct 27, 2021 • 1 new comment

• Keras 1D CNN always predicts the same result even if accuracy is high on training set

  #14376 commented on Oct 27, 2021 • 1 new comment

• Keras "index_array" referenced before assignment

  #14371 commented on Oct 27, 2021 • 1 new comment

• load_model - the two structures don't have the same nested structure

  #14345 commented on Oct 28, 2021 • 1 new comment

• input declaration issue with Embedding layer

  #14639 commented on Oct 28, 2021 • 1 new comment

• Improvement for tf.keras.utils.get_file

  #15540 commented on Oct 28, 2021 • 1 new comment

• TypeError in VQ-VAE example

  #15536 commented on Oct 29, 2021 • 1 new comment

• [tf2.6] Loaded subclassed-Model fails to be retrained

  #15261 commented on Oct 29, 2021 • 1 new comment

• Stateful LSTM gives different predictions for different input shapes

  #15232 commented on Oct 29, 2021 • 1 new comment

• TypeError: Parameter to MergeFrom() must be instance of same class: expected tensorflow.TensorShapeProto got tensorflow.TensorShapeProto.

  #15484 commented on Oct 30, 2021 • 1 new comment

• Changing the shape of given convolutional or dense layer

  #13932 commented on Oct 30, 2021 • 1 new comment

• Keras model.fit is displaying a big error, Unsolvable

  #14502 commented on Oct 30, 2021 • 1 new comment

• How to provide validation data for multitask classifier?

  #14622 commented on Oct 23, 2021 • 0 new comments

• EfficientNet transfer learning and greyscale images

  #14373 commented on Oct 23, 2021 • 0 new comments

• ValueError: Shapes (None, None) and (None, 7, 7, 67) are incompatible .. What's wrong with my code?

  #14341 commented on Oct 23, 2021 • 0 new comments

• Problem for passing integer seed value for tensorflow.keras.layers.experimental.preprocessing functions in .map function

  #15314 commented on Oct 24, 2021 • 0 new comments

• Performance of a model loading the same weights changes depending on the version of tensorflow

  #15513 commented on Oct 25, 2021 • 0 new comments

• load_model() does not return when layer of model was renamed

  #14356 commented on Oct 27, 2021 • 0 new comments

• DetermisticRandomTestTool - This tool is used to make random number generation semantics match between TF1.x graphs/sessions and eager execution.

  #15517 commented on Oct 29, 2021 • 0 new comments