GitLab

Replace per_device with per_replica and PerDevice with PerReplica, because the PerDevice concept was renamed and doesn't exist anymore. (#6693)

* Replace per_device with per_replica and PerDevice with PerReplica, because the PerReplica concept was renamed and doesn't exist anymore.

* bug fix

* bug fix

* Add benchmarks with the --cloning flag to Resnet and NFC.

* Renamed cloning to clone_model_in_keras_dist_strat. Dropped a few tests that aren't essential.

* Fixed up the formatting after re-naming the flag to a much longer  name.  Thanks, lint.
* Fixed the lint error in nfc_common.py

* Combined imagenet and cifar-10 benchmarks

* Comments and epochs_between_evals.

* Added tuned tests and cleaned up benchmark flags

* Fix names.

* Return results and add images/sec hook.

* updated doc strings for return values.

* 128 to 256 batch for FP16 test

* added more doc strings to fix lint.

* Add num_packs flag for MirroredStrategy's cross device ops.

* fix parens

* Fix lint errors and make all_reduce_alg more robust.

* Set default num_packs to 1

Reason: test failures because contrib is not available in V2

This reverts commit 325dd761.

* Added none check for output_dir

* Change double quote to single

* Introduce a short sleep before ds.prefetch in tf.data.

* Further limit dataset threads to reduce CPU contention

* Tuned dataset sleep time

* Rename dataset sleep flag; enable it only for Keras Graph mode

* Small word tweak

* Few more tweaks

Added additional information on using the `SavedModel` for prediction purposes.

* before moving test cases to the base class

* Added tests for keras cifar and keras imagenet

* fix cifar10_test

* add blank lines

* fix lint errors

* fix lint

* Resolve comments

* Modified two resnet keras tests

* Tests passed

* Remove keras_test_base

* Remove gpu from the no-dist tests

* Revert "Set input layer `batch_size` in multi-replica mode (#6578)"

This reverts commit f1a59682.

* Rename variables.

* Update resnet_run_loop.py

* Update resnet_run_loop.py

* Update resnet_run_loop.py

* Update resnet_run_loop.py

* Update resnet_run_loop.py

* Revert "Revert " Ensure static shapes when enabling XLA in Resnet Keras model (#6508)" (#6517)"

This reverts commit cc9eef76.

* Set `batch_size` to keras.Input in non-eager mode.

Eager mode currently has OOM problem.

* Add comments for enable_eager flag.

* Always set drop_remainder=True.

* Only set drop_remainder=True for XLA.

* Refactored ResNet code and added additional architectures.

* Added numerical layer names instead of alphabetical.

* Change dash to underscore.

* Corrected return statement.

* Use conv_strides argument.

* Set classes=10

* Use partial to reduce code duplication.

With trivial model, it improves the data input pipeline throughput from 12.5K to 15K on a DGX1 V100 machine.

* add ds support for ncf

* remove comments for in_top_k

* avoid expanding the input layers

* resolve comments and fix lint

* Added some comments in code and fix lint

* fix lint

* add some documentation

* add tensorflow imports

* Added main function to the two keras resent benchmarks

* remove asserts

Forgot to add the `dtype='fp16'` flag in previous CL...

* Add profiler callback for Keras models

* Update build stats to identify time callback by type

* Add warning message when both TensorBoard and profiler callbacks are used

* Add benchmarks for dynamic loss scaling.

I add one convergence benchmark and five performance benchmarks.

* Fix lint error

This should give better performance in tweaked benchmarks.

Reason: break 1-gpu nightly test.

This reverts commit 371645fc.

Don't pass `batch_size` to keras.layers.Input in DS multi-replica case. There is currently a bug in Keras side which will cause a batch size incompatible error.

* Update resnet_model.py

* Ensure static shapes when enabling XLA.

* Define `drop_remainder` as a variable.

* Handles per_replica_batch_size in non-XLA mode

* Remove trailing whitespace.

* Add tweaked test for fp16 nonXLA

* Add dummy test for dataset warmup