Fix the tpu related broken link especially for imagenet_to_gcs.py (#18126)

diff --git a/tensorflow/contrib/tpu/tpu_estimator.md b/tensorflow/contrib/tpu/tpu_estimator.md
index 4ef8f9e..639e708 100644 (file)
--- a/tensorflow/contrib/tpu/tpu_estimator.md
+++ b/tensorflow/contrib/tpu/tpu_estimator.md
@@ -172,7 +172,7 @@ It is always recommended to port a small, simple model first to make sure that
 you are familiar with the basic concepts of `TPUEstimator` and test end-to-end
 behavior. Once your simple model runs, gradually add more functionality.
 In addition, there are several sample models, available at
-[github.com/tensorflow/tpu-demos](https://github.com/tensorflow/tpu-demos).
+[github.com/tensorflow/tpu](https://github.com/tensorflow/tpu).
 
 To convert your code from the vanilla `Estimator` class to use TPUs, change the
 following (note some of the details may change over time):

diff --git a/tensorflow/docs_src/programmers_guide/using_tpu.md b/tensorflow/docs_src/programmers_guide/using_tpu.md
index a9c2cb3..cb0d86f 100644 (file)
--- a/tensorflow/docs_src/programmers_guide/using_tpu.md
+++ b/tensorflow/docs_src/programmers_guide/using_tpu.md
@@ -11,7 +11,7 @@ This doc is aimed at users who:
   using an existing model.
 * Have, perhaps, skimmed the code of an example TPU model
   [[1]](https://github.com/tensorflow/models/blob/master/official/mnist/mnist_tpu.py)
-  [[2]](https://github.com/tensorflow/tpu-demos/tree/master/cloud_tpu/models).
+  [[2]](https://github.com/tensorflow/tpu/tree/master/models).
 * Are interested in porting an existing `Estimator` model to
   run on Cloud TPUs
 
@@ -288,7 +288,7 @@ If shape inference has failed, but the shape is known it is possible to
 impose the correct shape using `tf.set_shape()`. 
 
 In the example below the shape
-inference algorithm fails, but it is corrected using `set_shape`:
+inference algorithm fails, but it is correctly using `set_shape`:
 
 ```
 >>> x = tf.zeros(tf.constant([1,2,3])+1)
@@ -371,10 +371,10 @@ in bytes. A minimum of a few MB (`buffer_size=8*1024*1024`) is recommended so
 that data is available when needed.
 
 The TPU-demos repo includes
-[a script](https://github.com/tensorflow/tpu-demos/blob/master/cloud_tpu/datasets/imagenet_to_gcs.py)
+[a script](https://github.com/tensorflow/tpu/blob/master/tools/datasets/imagenet_to_gcs.py)
 for downloading the imagenet dataset and converting it to an appropriate format.
 This together with the imagenet
-[models](https://github.com/tensorflow/tpu-demos/tree/master/cloud_tpu/models)
+[models](https://github.com/tensorflow/tpu/tree/master/models)
 included in the repo demonstrate all of these best-practices.
 
 
@@ -387,7 +387,7 @@ For details on how to actually set up and run a Cloud TPU see:
 This document is by no means exhaustive. The best source of more detail on how
 to make a Cloud TPU compatible model are the example models published in:
 
- * The [TPU Demos Repository.](https://github.com/tensorflow/tpu-demos/)
+ * The [TPU Demos Repository.](https://github.com/tensorflow/tpu)
 
 For more information about tuning TensorFlow code for performance see: