contrib: minor spelling tweaks

packages:
  data
  training
  tensor_forest

11 files changed, 13 insertions, 13 deletions

diff --git a/tensorflow/contrib/tensor_forest/client/random_forest.py b/tensorflow/contrib/tensor_forest/client/random_forest.py
index 4abcc20ed3..35e8c92aba 100644
--- a/tensorflow/contrib/tensor_forest/client/random_forest.py
+++ b/tensorflow/contrib/tensor_forest/client/random_forest.py
@@ -399,7 +399,7 @@ def get_combined_model_fn(model_fns):
   training ops: tf.group them.
   loss: average them.
   predictions: concat probabilities such that predictions[*][0-C1] are the
-    probablities for output 1 (where C1 is the number of classes in output 1),
+    probabilities for output 1 (where C1 is the number of classes in output 1),
     predictions[*][C1-(C1+C2)] are the probabilities for output 2 (where C2
     is the number of classes in output 2), etc.  Also stack predictions such
     that predictions[i][j] is the class prediction for example i and output j.
diff --git a/tensorflow/contrib/tensor_forest/hybrid/core/ops/hard_routing_function_op.cc b/tensorflow/contrib/tensor_forest/hybrid/core/ops/hard_routing_function_op.cc
index cf0db788a4..06bfe871fd 100644
--- a/tensorflow/contrib/tensor_forest/hybrid/core/ops/hard_routing_function_op.cc
+++ b/tensorflow/contrib/tensor_forest/hybrid/core/ops/hard_routing_function_op.cc
@@ -80,7 +80,7 @@ REGISTER_OP("HardRoutingFunction")
    regression model that translates from node features to
    probabilities.
 
-  path_probility: `path_probability[i]` gives the probability of reaching each
+  path_probability: `path_probability[i]` gives the probability of reaching each
    node in `path[i]`.
   path: `path[i][j]` gives the jth node in the path taken by the ith data
    instance.
diff --git a/tensorflow/contrib/tensor_forest/hybrid/core/ops/stochastic_hard_routing_function_op.cc b/tensorflow/contrib/tensor_forest/hybrid/core/ops/stochastic_hard_routing_function_op.cc
index c9df09bfda..1a055756c0 100644
--- a/tensorflow/contrib/tensor_forest/hybrid/core/ops/stochastic_hard_routing_function_op.cc
+++ b/tensorflow/contrib/tensor_forest/hybrid/core/ops/stochastic_hard_routing_function_op.cc
@@ -85,7 +85,7 @@ REGISTER_OP("StochasticHardRoutingFunction")
    regression model that translates from node features to
    probabilities.
 
-  path_probility: `path_probability[i]` gives the probability of reaching each
+  path_probability: `path_probability[i]` gives the probability of reaching each
    node in `path[i]`.
   path: `path[i][j]` gives the jth node in the path taken by the ith data
    instance.
diff --git a/tensorflow/contrib/tensor_forest/hybrid/core/ops/stochastic_hard_routing_gradient_op.cc b/tensorflow/contrib/tensor_forest/hybrid/core/ops/stochastic_hard_routing_gradient_op.cc
index b0d8b832b5..7d092bbc24 100644
--- a/tensorflow/contrib/tensor_forest/hybrid/core/ops/stochastic_hard_routing_gradient_op.cc
+++ b/tensorflow/contrib/tensor_forest/hybrid/core/ops/stochastic_hard_routing_gradient_op.cc
@@ -81,7 +81,7 @@ REGISTER_OP("StochasticHardRoutingGradient")
   tree_biases: `tree_biases[i]` gives the bias of the logistic
    regression model that translates from node features to
    probabilities.
-  path_probility: `path_probability[i]` gives the probability of reaching each
+  path_probability: `path_probability[i]` gives the probability of reaching each
    node in `path[i]`.
   path: `path[i][j]` gives the jth node in the path taken by the ith data
    instance.
diff --git a/tensorflow/contrib/tensor_forest/kernels/tree_utils.cc b/tensorflow/contrib/tensor_forest/kernels/tree_utils.cc
index 44997ec5d6..cefcc96051 100644
--- a/tensorflow/contrib/tensor_forest/kernels/tree_utils.cc
+++ b/tensorflow/contrib/tensor_forest/kernels/tree_utils.cc
@@ -421,7 +421,7 @@ double getChebyshevEpsilon(const std::vector<float>& mu1,
                            const std::vector<float>& mu2) {
   // Math time!!
   // We are trying to minimize d = |mu1 - x|^2 + |mu2 - y|^2 over the surface.
-  // Using Langrange multipliers, we get
+  // Using Lagrange multipliers, we get
   //   partial d / partial x = -2 mu1 + 2 x = lambda_1 1 + 2 lambda_3 x
   //   partial d / partial y = -2 mu2 + 2 y = lambda_2 1 - 2 lambda_3 y
   // or
@@ -485,7 +485,7 @@ double getChebyshevEpsilon(const std::vector<float>& mu1,
   }
 
   double sdiscrim = sqrt(discrim);
-  // TODO(thomaswc): Analyze whetever one of these is always closer.
+  // TODO(thomaswc): Analyze whatever one of these is always closer.
   double v1 = (-b + sdiscrim) / (2 * a);
   double v2 = (-b - sdiscrim) / (2 * a);
   double dist1 = getDistanceFromLambda3(v1, mu1, mu2);
diff --git a/tensorflow/contrib/tensor_forest/kernels/tree_utils.h b/tensorflow/contrib/tensor_forest/kernels/tree_utils.h
index edbac67006..03aab1b61e 100644
--- a/tensorflow/contrib/tensor_forest/kernels/tree_utils.h
+++ b/tensorflow/contrib/tensor_forest/kernels/tree_utils.h
@@ -123,7 +123,7 @@ bool BestSplitDominatesRegression(const Tensor& total_sums,
                                   const Tensor& split_squares,
                                   int32 accumulator);
 
-// Performs booststrap_samples bootstrap samples of the best split's class
+// Performs bootstrap_samples bootstrap samples of the best split's class
 // counts and the second best splits's class counts, and returns true if at
 // least dominate_fraction of the time, the former has a better (lower)
 // Gini impurity.  Does not take over ownership of *rand.
diff --git a/tensorflow/contrib/tensor_forest/kernels/v4/decision-tree-resource.h b/tensorflow/contrib/tensor_forest/kernels/v4/decision-tree-resource.h
index 328af28725..d3edb43733 100644
--- a/tensorflow/contrib/tensor_forest/kernels/v4/decision-tree-resource.h
+++ b/tensorflow/contrib/tensor_forest/kernels/v4/decision-tree-resource.h
@@ -60,7 +60,7 @@ class DecisionTreeResource : public ResourceBase {
   mutex* get_mutex() { return &mu_; }
 
   // Return the TreeNode for the leaf that the example ends up at according
-  // to decsion_tree_. Also fill in that leaf's depth if it isn't nullptr.
+  // to decision_tree_. Also fill in that leaf's depth if it isn't nullptr.
   int32 TraverseTree(const std::unique_ptr<TensorDataSet>& input_data,
                      int example, int32* depth, TreePath* path) const;
 
diff --git a/tensorflow/contrib/tensor_forest/kernels/v4/decision_node_evaluator.h b/tensorflow/contrib/tensor_forest/kernels/v4/decision_node_evaluator.h
index bf2b2aaa3c..3db351c328 100644
--- a/tensorflow/contrib/tensor_forest/kernels/v4/decision_node_evaluator.h
+++ b/tensorflow/contrib/tensor_forest/kernels/v4/decision_node_evaluator.h
@@ -60,7 +60,7 @@ class InequalityDecisionNodeEvaluator : public BinaryDecisionNodeEvaluator {
   bool include_equals_;
 };
 
-// Evalutor for splits with multiple weighted features.
+// Evaluator for splits with multiple weighted features.
 class ObliqueInequalityDecisionNodeEvaluator
     : public BinaryDecisionNodeEvaluator {
  public:
diff --git a/tensorflow/contrib/tensor_forest/ops/model_ops.cc b/tensorflow/contrib/tensor_forest/ops/model_ops.cc
index 3099cccdf8..98124d519c 100644
--- a/tensorflow/contrib/tensor_forest/ops/model_ops.cc
+++ b/tensorflow/contrib/tensor_forest/ops/model_ops.cc
@@ -165,7 +165,7 @@ tree_handle: The handle to the tree.
 leaf_ids: `leaf_ids[i]` is the leaf id for input i.
 input_labels: The training batch's labels as a 1 or 2-d tensor.
   'input_labels[i][j]' gives the j-th label/target for the i-th input.
-input_weights: The training batch's eample weights as a 1-d tensor.
+input_weights: The training batch's weights as a 1-d tensor.
   'input_weights[i]' gives the weight for the i-th input.
 )doc");
 
diff --git a/tensorflow/contrib/tensor_forest/ops/stats_ops.cc b/tensorflow/contrib/tensor_forest/ops/stats_ops.cc
index e8b5c5d8a6..be0a11546d 100644
--- a/tensorflow/contrib/tensor_forest/ops/stats_ops.cc
+++ b/tensorflow/contrib/tensor_forest/ops/stats_ops.cc
@@ -83,7 +83,7 @@ Grows the tree for finished nodes and allocates waiting nodes.
 params: A serialized TensorForestParams proto.
 tree_handle: The handle to the tree.
 stats_handle: The handle to the stats.
-finshed_nodes: A 1-d Tensor of finished node ids from ProcessInput.
+finished_nodes: A 1-d Tensor of finished node ids from ProcessInput.
 )doc");
 
 REGISTER_OP("ProcessInputV4")
@@ -119,7 +119,7 @@ sparse_input_values: The values tensor from the SparseTensor input.
 sparse_input_shape: The shape tensor from the SparseTensor input.
 input_labels: The training batch's labels as a 1 or 2-d tensor.
   'input_labels[i][j]' gives the j-th label/target for the i-th input.
-input_weights: The training batch's eample weights as a 1-d tensor.
+input_weights: The training batch's weights as a 1-d tensor.
   'input_weights[i]' gives the weight for the i-th input.
 finished_nodes: A 1-d tensor of node ids that have finished and are ready to
   grow.
diff --git a/tensorflow/contrib/tensor_forest/python/tensor_forest.py b/tensorflow/contrib/tensor_forest/python/tensor_forest.py
index 3650b5d52f..b9bcbb170b 100644
--- a/tensorflow/contrib/tensor_forest/python/tensor_forest.py
+++ b/tensorflow/contrib/tensor_forest/python/tensor_forest.py
@@ -212,7 +212,7 @@ class ForestHParams(object):
     self.regression = getattr(self, 'regression', False)
 
     # Num_outputs is the actual number of outputs (a single prediction for
-    # classification, a N-dimenensional point for regression).
+    # classification, a N-dimensional point for regression).
     self.num_outputs = self.num_classes if self.regression else 1
 
     # Add an extra column to classes for storing counts, which is needed for