GitLab

Fix Issue : Change exporter.py in object_detection

[mnist]: Training and evaluation with TPUs.

Incorporated the suggested changes.

[blog_estimators_dataset.py] bug fix for failing demo

Fix spacing in BUILD file comments.

Open source release of Brain Coder.

update supported tf versions for object detection on gcloud.

This is a step towards merging the example in
https://github.com/tensorflow/tpu-demos/tree/master/cloud_tpu/models/mnist
with this repository, so we have a single model definition for
training across CPU/GPU/eager execution/TPU.

The change to dataset.py is so that the raw data can be read
from cloud storage systems (like GCS and S3).

Fix demo_inference to properly normalize input.

Use temp variable for predictions.

Before the fix the demo_inference.py used batch_norm and it did the
normalization of input image implicitly. If at inference time the
batch_norm was disabled the inference produced incorrect results.
This fix does the proper input image normalization and disables the batch_norm
at inference time.

Revert "Flags"

Change in Object Dectection Model. Change parameter name given to RewriterConfig Proto. optimize_tensor_layout was changed to layout_optimizer in current tf version.

Update Dockerfile with new bazel version.

Update README.md

Fix AudioSet url

[mnist]: Use FixedLengthRecordDataset

- Prior to this change, the use of tf.data.Dataset essentially embedded
  the entire training/evaluation dataset into the graph as a constant,
  leading to unnecessarily humungous graphs (Fixes #3017)
- Also, use batching on the evaluation dataset to allow
  evaluation on GPUs that cannot fit the entire evaluation dataset in
  memory (Fixes #3046)

Update README.md

Fix typo: "the most is mostly converged" vs "the model is mostly converged"

compatible with tensorflow 1.4.1

Redirect users from cifar10 to cifar10_estimator

Make the redirect to slim notice more prominent

[mnist]: Make flags function scoped (not module scoped).

This will make it easier to share the model definition with
eager execution and TPU demos without any side effects of running
unnecessary code on module import.

[mnist]: Updates