diff options
author | Jacques Pienaar <jpienaar@google.com> | 2018-03-21 12:07:51 -0700 |
---|---|---|
committer | TensorFlower Gardener <gardener@tensorflow.org> | 2018-03-21 12:10:30 -0700 |
commit | 2d0531d72c7dcbb0e149cafdd3a16ee8c3ff357a (patch) | |
tree | 1179ecdd684d10c6549f85aa95f33dd79463a093 /tensorflow/contrib/gan | |
parent | cbede3ea7574b36f429710bc08617d08455bcc21 (diff) |
Merge changes from github.
PiperOrigin-RevId: 189945839
Diffstat (limited to 'tensorflow/contrib/gan')
5 files changed, 10 insertions, 10 deletions
diff --git a/tensorflow/contrib/gan/python/eval/python/classifier_metrics_impl.py b/tensorflow/contrib/gan/python/eval/python/classifier_metrics_impl.py index 7e86d10b64..47e51415fd 100644 --- a/tensorflow/contrib/gan/python/eval/python/classifier_metrics_impl.py +++ b/tensorflow/contrib/gan/python/eval/python/classifier_metrics_impl.py @@ -321,7 +321,7 @@ def classifier_score(images, classifier_fn, num_batches=1): NOTE: This function consumes images, computes their logits, and then computes the classifier score. If you would like to precompute many logits for - large batches, use clasifier_score_from_logits(), which this method also + large batches, use classifier_score_from_logits(), which this method also uses. Args: @@ -454,7 +454,7 @@ def frechet_classifier_distance(real_images, This technique is described in detail in https://arxiv.org/abs/1706.08500. Given two Gaussian distribution with means m and m_w and covariance matrices - C and C_w, this function calcuates + C and C_w, this function calculates |m - m_w|^2 + Tr(C + C_w - 2(C * C_w)^(1/2)) @@ -467,7 +467,7 @@ def frechet_classifier_distance(real_images, Frechet distance is biased. It is more biased for small sample sizes. (e.g. even if the two distributions are the same, for a small sample size, the expected Frechet distance is large). It is important to use the same - sample size to compute frechet classifier distance when comparing two + sample size to compute Frechet classifier distance when comparing two generative models. NOTE: This function consumes images, computes their activations, and then @@ -659,7 +659,7 @@ def frechet_classifier_distance_from_activations(real_activations, This technique is described in detail in https://arxiv.org/abs/1706.08500. Given two Gaussian distribution with means m and m_w and covariance matrices - C and C_w, this function calcuates + C and C_w, this function calculates |m - m_w|^2 + Tr(C + C_w - 2(C * C_w)^(1/2)) diff --git a/tensorflow/contrib/gan/python/eval/python/sliced_wasserstein_impl.py b/tensorflow/contrib/gan/python/eval/python/sliced_wasserstein_impl.py index 9bebcacbe4..4b10bc0f8e 100644 --- a/tensorflow/contrib/gan/python/eval/python/sliced_wasserstein_impl.py +++ b/tensorflow/contrib/gan/python/eval/python/sliced_wasserstein_impl.py @@ -212,7 +212,7 @@ def sliced_wasserstein_distance(real_images, Args: real_images: (tensor) Real images (batch, height, width, channels). fake_images: (tensor) Fake images (batch, height, width, channels). - resolution_min: (int) Minimum resolution for the Laplacion pyramid. + resolution_min: (int) Minimum resolution for the Laplacian pyramid. patches_per_image: (int) Number of patches to extract per image per Laplacian level. patch_size: (int) Width of a square patch. @@ -221,7 +221,7 @@ def sliced_wasserstein_distance(real_images, use_svd: experimental method to compute a more accurate distance. Returns: List of tuples (distance_real, distance_fake) for each level of the - Laplacian pyramid from the highest resoluion to the lowest. + Laplacian pyramid from the highest resolution to the lowest. distance_real is the Wasserstein distance between real images distance_fake is the Wasserstein distance between real and fake images. Raises: diff --git a/tensorflow/contrib/gan/python/features/python/conditioning_utils_impl.py b/tensorflow/contrib/gan/python/features/python/conditioning_utils_impl.py index cd31c62667..e2594faf85 100644 --- a/tensorflow/contrib/gan/python/features/python/conditioning_utils_impl.py +++ b/tensorflow/contrib/gan/python/features/python/conditioning_utils_impl.py @@ -12,7 +12,7 @@ # See the License for the specific language governing permissions and # limitations under the License. # ============================================================================== -"""Miscellanous utilities for TFGAN code and examples. +"""Miscellaneous utilities for TFGAN code and examples. Includes: 1) Conditioning the value of a Tensor, based on techniques from diff --git a/tensorflow/contrib/gan/python/features/python/random_tensor_pool_impl.py b/tensorflow/contrib/gan/python/features/python/random_tensor_pool_impl.py index 4cfae0de44..9e4ec59e70 100644 --- a/tensorflow/contrib/gan/python/features/python/random_tensor_pool_impl.py +++ b/tensorflow/contrib/gan/python/features/python/random_tensor_pool_impl.py @@ -17,7 +17,7 @@ We use this to keep a history of values created by a generator, such that a discriminator can randomly be trained on some older samples, not just the current one. This can help to not let the discriminator get too far ahead of the -generator and also to keep the system from oscilating, if the discriminator +generator and also to keep the system from oscillating, if the discriminator forgets too fast what past samples from the generator looked like. See the following papers for more details. @@ -97,7 +97,7 @@ def tensor_pool(input_values, dtypes=[v.dtype for v in input_values], shapes=None) - # In pseudeo code this code does the following: + # In pseudo code this code does the following: # if not pool_full: # enqueue(input_values) # return input_values diff --git a/tensorflow/contrib/gan/python/features/python/virtual_batchnorm_test.py b/tensorflow/contrib/gan/python/features/python/virtual_batchnorm_test.py index 845f89827b..2fe06a2872 100644 --- a/tensorflow/contrib/gan/python/features/python/virtual_batchnorm_test.py +++ b/tensorflow/contrib/gan/python/features/python/virtual_batchnorm_test.py @@ -148,7 +148,7 @@ class VirtualBatchnormTest(test.TestCase): self.assertAllClose(bn_np[i, ...], vb_np) def test_minibatch_independent(self): - """Test that virtual batch normalized exampels are independent. + """Test that virtual batch normalized examples are independent. Unlike batch normalization, virtual batch normalization has the property that the virtual batch normalized value of an example is independent of the |