Qianli Scott Zhu

Keras Tuner stopped at the end of first epoch when train on TPU

20

This is discovered during test of https://www.kaggle.com/kivlichangoogle/jigsaw-multilingual-getting-started. Note that this issue only happens when TPU is enabled for the notebook. When Keras tuner is added, the search process will fail...

Update current FPN for the extra level support. This is an extension that is used by the Retina-net application

3

# What does this PR do? Update current FPN for the extra level support. This is an extension that is used by the Retina-net application. We could use this as...

Add the implementation for DeepLab v3 based on TFM implementation (as an experiment)

9

# What does this PR do? Not intended to merge for now. Add the implementation for DeepLab v3 based on TFM implementation (as an experiment) The current class is a...

Add MaskRCNN model as a test for the wrapper approach with TF model garden code

3

# Note that this PR is more for a demo and not intend to be submitted at the moment. # What does this PR do? Add MaskRCNN model as a...

[Memory optimization] Reduce the memory usage for preset() API with load the weights.

1

**Is your feature request related to a problem? Please describe.** This is a request forwarded from @martin-gorner based on the user feedback. Currently when creating model from preset() API with...

@fchollet, this is the draft of the OPT model inferencing with Keras distribution API (I can add the finetune parts later). Do u still have the instructions to convert the...

[Distribution] Standardize the device API.

3

1. Unify the string representation for the devices among all the backends, e.g. all devices should be lower case string, and standardized on 'cpu', 'gpu', 'tpu' with ':0' or ':1'...

Nit fix the missing include of thread

5

Notice this in our internal build system. `std::this_thread::yield()` would require the include of thread.