GitLab

Fix github issue #3269 where the accuracy is wrongly underestimated for binary classification and build issue #2784

Move gen_data and gen_vocab to parent directory to prevent import madness in the open-source code. Remove unnecessary bazel dependency and documentation in README.

PiperOrigin-RevId: 186638059

Fix license text

Seems like it was changed accidentally as it is different than any of the other files.

PiperOrigin-RevId: 186541910

Internal changes for object detection.

Fix PythonPath dependencies

PiperOrigin-RevId: 186265033

Merging in improvements and fixes to adversarial_text

Previously, code block one attempted to import `from object_detection.utils` before the `object_detection` directory was added to the PythonPath via `sys.path.append("..")`.

This generated a `ModuleNotFoundError`.

Accordingly, append `".."` to the PythonPath before attempting to import from `object_detection` as a module, instead of after.

PiperOrigin-RevId: 185783053

Replace tf.contrib.framework.get_or_create_global_step() with tf.train.get_or_create_global_step() in third_party/tensorflow_models

See https://www.tensorflow.org/api_docs/python/tf/contrib/framework/get_or_create_global_step and #2795

PiperOrigin-RevId: 176534973

PiperOrigin-RevId: 173414999

Add option for when there is only one label for classification, for speed. (The case for the standard datasets).

PiperOrigin-RevId: 172774825

PiperOrigin-RevId: 172665358

memory and fixes the OOM errors.

PiperOrigin-RevId: 172374935

Change logit and activation shapes to follow standard convention with batch size in the first dimension.

PiperOrigin-RevId: 172172495

PiperOrigin-RevId: 171498799

MNIST: Eager execution.

Add an example showing how to train the MNIST model with eager execution
enabled.

(This change requires changes to TensorFlow made after
the 1.6 release branch was cut, i.e., will require a build
from source or TensorFlow 1.7+)

* Refining preprocessing, part 1

* Refinements to preprocessing resulting from multi-GPU tests

* Reviving one-hot labels

* Reviving one-hot labels

* Fixing label shapes

* Adding random flip back in

* Reverting unnecessary linting of test file

* Respond to CR

* Respond to CR

* Respond to CR

Small polishing changes to DELF code.

Sync with master.

PiperOrigin-RevId: 185555440

2. Change tfexample_decoder in slim/objection_detection to accept different JPEG decompression method.
Defaults to ""/None which maps to a system-specific default. Currently valid values are ["INTEGER_FAST", "INTEGER_ACCURATE"]. The hint may be ignored (e.g., the internal jpeg library changes to a version that does not have that specific option.)

PiperOrigin-RevId: 185486653

PiperOrigin-RevId: 185470191

PiperOrigin-RevId: 185425302

PiperOrigin-RevId: 185315101

PiperOrigin-RevId: 185229737

Add encode_background_as_zeros option to the SSDMetaArch class --- now clients have the option of encoding background targets as an all zeros vector or a one-hot vector with the 0th dimension corresponding to a background prediction.

PiperOrigin-RevId: 185228281