diff options
author | 2016-07-22 08:20:00 -0800 | |
---|---|---|
committer | 2016-07-22 09:33:16 -0700 | |
commit | df7d42f3c34f0aa3dc5ddc2c175366b0f8a4a802 (patch) | |
tree | 93cb608ca5d900e075debd4d9d537718a84f5a80 | |
parent | 2a772ed74613d8842de2efb10282830c9b368174 (diff) |
Update generated Python Op docs.
Change: 128180221
13 files changed, 804 insertions, 0 deletions
diff --git a/tensorflow/g3doc/api_docs/python/contrib.layers.md b/tensorflow/g3doc/api_docs/python/contrib.layers.md index 42afb94293..914eb0f581 100644 --- a/tensorflow/g3doc/api_docs/python/contrib.layers.md +++ b/tensorflow/g3doc/api_docs/python/contrib.layers.md @@ -13,6 +13,85 @@ common machine learning algorithms. - - - +### `tf.contrib.layers.avg_pool2d(*args, **kwargs)` {#avg_pool2d} + +Adds a Avg Pooling op. + +It is assumed by the wrapper that the pooling is only done per image and not +in depth or batch. + +##### Args: + + +* <b>`inputs`</b>: a tensor of size [batch_size, height, width, depth]. +* <b>`kernel_size`</b>: a list of length 2: [kernel_height, kernel_width] of the + pooling kernel over which the op is computed. Can be an int if both + values are the same. +* <b>`stride`</b>: a list of length 2: [stride_height, stride_width]. + Can be an int if both strides are the same. Note that presently + both strides must have the same value. +* <b>`padding`</b>: the padding method, either 'VALID' or 'SAME'. +* <b>`outputs_collections`</b>: collection to add the outputs. +* <b>`scope`</b>: Optional scope for op_scope. + +##### Returns: + + a tensor representing the results of the pooling operation. + + +- - - + +### `tf.contrib.layers.batch_norm(*args, **kwargs)` {#batch_norm} + +Adds a Batch Normalization layer from http://arxiv.org/abs/1502.03167. + + "Batch Normalization: Accelerating Deep Network Training by Reducing + Internal Covariate Shift" + + Sergey Ioffe, Christian Szegedy + +Can be used as a normalizer function for conv2d and fully_connected. + +##### Args: + + +* <b>`inputs`</b>: a tensor of size `[batch_size, height, width, channels]` + or `[batch_size, channels]`. +* <b>`decay`</b>: decay for the moving average. +* <b>`center`</b>: If True, subtract `beta`. If False, `beta` is ignored. +* <b>`scale`</b>: If True, multiply by `gamma`. If False, `gamma` is + not used. When the next layer is linear (also e.g. `nn.relu`), this can be + disabled since the scaling can be done by the next layer. +* <b>`epsilon`</b>: small float added to variance to avoid dividing by zero. +* <b>`activation_fn`</b>: Optional activation function. +* <b>`updates_collections`</b>: collections to collect the update ops for computation. + If None, a control dependency would be added to make sure the updates are + computed. +* <b>`is_training`</b>: whether or not the layer is in training mode. In training mode + it would accumulate the statistics of the moments into `moving_mean` and + `moving_variance` using an exponential moving average with the given + `decay`. When it is not in training mode then it would use the values of + the `moving_mean` and the `moving_variance`. +* <b>`reuse`</b>: whether or not the layer and its variables should be reused. To be + able to reuse the layer scope must be given. +* <b>`variables_collections`</b>: optional collections for the variables. +* <b>`outputs_collections`</b>: collections to add the outputs. +* <b>`trainable`</b>: If `True` also add variables to the graph collection + `GraphKeys.TRAINABLE_VARIABLES` (see tf.Variable). +* <b>`scope`</b>: Optional scope for `variable_op_scope`. + +##### Returns: + + A `Tensor` representing the output of the operation. + +##### Raises: + + +* <b>`ValueError`</b>: if rank or last dimension of `inputs` is undefined. + + +- - - + ### `tf.contrib.layers.convolution2d(*args, **kwargs)` {#convolution2d} Adds a 2D convolution followed by an optional batch_norm layer. @@ -72,6 +151,129 @@ greater than one. - - - +### `tf.contrib.layers.convolution2d_in_plane(*args, **kwargs)` {#convolution2d_in_plane} + +Performs the same in-plane convolution to each channel independently. + +This is useful for performing various simple channel-independent convolution +operations such as image gradients: + + image = tf.constant(..., shape=(16, 240, 320, 3)) + vert_gradients = layers.conv2d_in_plane(image, + kernel=[1, -1], + kernel_size=[2, 1]) + horz_gradients = layers.conv2d_in_plane(image, + kernel=[1, -1], + kernel_size=[1, 2]) + +##### Args: + + +* <b>`inputs`</b>: a 4-D tensor with dimensions [batch_size, height, width, channels]. +* <b>`kernel_size`</b>: a list of length 2 holding the [kernel_height, kernel_width] of + of the pooling. Can be an int if both values are the same. +* <b>`stride`</b>: a list of length 2 `[stride_height, stride_width]`. + Can be an int if both strides are the same. Note that presently + both strides must have the same value. +* <b>`padding`</b>: the padding type to use, either 'SAME' or 'VALID'. +* <b>`activation_fn`</b>: activation function. +* <b>`normalizer_fn`</b>: normalization function to use instead of `biases`. If + `normalize_fn` is provided then `biases_initializer` and + `biases_regularizer` are ignored and `biases` are not created nor added. +* <b>`normalizer_params`</b>: normalization function parameters. +* <b>`weights_initializer`</b>: An initializer for the weights. +* <b>`weights_regularizer`</b>: Optional regularizer for the weights. +* <b>`biases_initializer`</b>: An initializer for the biases. If None skip biases. +* <b>`biases_regularizer`</b>: Optional regularizer for the biases. +* <b>`reuse`</b>: whether or not the layer and its variables should be reused. To be + able to reuse the layer scope must be given. +* <b>`variables_collections`</b>: optional list of collections for all the variables or + a dictionay containing a different list of collection per variable. +* <b>`outputs_collections`</b>: collection to add the outputs. +* <b>`trainable`</b>: If `True` also add variables to the graph collection + `GraphKeys.TRAINABLE_VARIABLES` (see tf.Variable). +* <b>`scope`</b>: Optional scope for `variable_op_scope`. + +##### Returns: + + A `Tensor` representing the output of the operation. + + +- - - + +### `tf.contrib.layers.convolution2d_transpose(*args, **kwargs)` {#convolution2d_transpose} + +Adds a convolution2d_transpose with an optional batch normalization layer. + +The function creates a variable called `weights`, representing the +kernel, that is convolved with the input. If `batch_norm_params` is `None`, a +second variable called 'biases' is added to the result of the operation. + +##### Args: + + +* <b>`inputs`</b>: a tensor of size [batch_size, height, width, channels]. +* <b>`num_outputs`</b>: integer, the number of output filters. +* <b>`kernel_size`</b>: a list of length 2 holding the [kernel_height, kernel_width] of + of the filters. Can be an int if both values are the same. +* <b>`stride`</b>: a list of length 2: [stride_height, stride_width]. + Can be an int if both strides are the same. Note that presently + both strides must have the same value. +* <b>`padding`</b>: one of 'VALID' or 'SAME'. +* <b>`activation_fn`</b>: activation function. +* <b>`normalizer_fn`</b>: normalization function to use instead of `biases`. If + `normalize_fn` is provided then `biases_initializer` and + `biases_regularizer` are ignored and `biases` are not created nor added. +* <b>`normalizer_params`</b>: normalization function parameters. +* <b>`weights_initializer`</b>: An initializer for the weights. +* <b>`weights_regularizer`</b>: Optional regularizer for the weights. +* <b>`biases_initializer`</b>: An initializer for the biases. If None skip biases. +* <b>`biases_regularizer`</b>: Optional regularizer for the biases. +* <b>`reuse`</b>: whether or not the layer and its variables should be reused. To be + able to reuse the layer scope must be given. +* <b>`variables_collections`</b>: optional list of collections for all the variables or + a dictionay containing a different list of collection per variable. +* <b>`outputs_collections`</b>: collection to add the outputs. +* <b>`trainable`</b>: whether or not the variables should be trainable or not. +* <b>`scope`</b>: Optional scope for variable_op_scope. + +##### Returns: + + a tensor representing the output of the operation. + +##### Raises: + + +* <b>`ValueError`</b>: if 'kernel_size' is not a list of length 2. + + +- - - + +### `tf.contrib.layers.flatten(*args, **kwargs)` {#flatten} + +Flattens the input while maintaining the batch_size. + + Assumes that the first dimension represents the batch. + +##### Args: + + +* <b>`inputs`</b>: a tensor of size [batch_size, ...]. +* <b>`outputs_collections`</b>: collection to add the outputs. +* <b>`scope`</b>: Optional scope for op_scope. + +##### Returns: + + a flattened tensor with shape [batch_size, k]. + +##### Raises: + + +* <b>`ValueError`</b>: if inputs.shape is wrong. + + +- - - + ### `tf.contrib.layers.fully_connected(*args, **kwargs)` {#fully_connected} Adds a fully connected layer. @@ -121,6 +323,217 @@ prior to the initial matrix multiply by `weights`. * <b>`ValueError`</b>: if x has rank less than 2 or if its last dimension is not set. +- - - + +### `tf.contrib.layers.max_pool2d(*args, **kwargs)` {#max_pool2d} + +Adds a Max Pooling op. + +It is assumed by the wrapper that the pooling is only done per image and not +in depth or batch. + +##### Args: + + +* <b>`inputs`</b>: a tensor of size [batch_size, height, width, depth]. +* <b>`kernel_size`</b>: a list of length 2: [kernel_height, kernel_width] of the + pooling kernel over which the op is computed. Can be an int if both + values are the same. +* <b>`stride`</b>: a list of length 2: [stride_height, stride_width]. + Can be an int if both strides are the same. Note that presently + both strides must have the same value. +* <b>`padding`</b>: the padding method, either 'VALID' or 'SAME'. +* <b>`outputs_collections`</b>: collection to add the outputs. +* <b>`scope`</b>: Optional scope for op_scope. + +##### Returns: + + a tensor representing the results of the pooling operation. + +##### Raises: + + +* <b>`ValueError`</b>: if 'kernel_size' is not a 2-D list + + +- - - + +### `tf.contrib.layers.one_hot_encoding(*args, **kwargs)` {#one_hot_encoding} + +Transform numeric labels into onehot_labels using tf.one_hot. + +##### Args: + + +* <b>`labels`</b>: [batch_size] target labels. +* <b>`num_classes`</b>: total number of classes. +* <b>`on_value`</b>: A scalar defining the on-value. +* <b>`off_value`</b>: A scalar defining the off-value. +* <b>`outputs_collections`</b>: collection to add the outputs. +* <b>`scope`</b>: Optional scope for op_scope. + +##### Returns: + + one hot encoding of the labels. + + +- - - + +### `tf.contrib.layers.repeat(inputs, repetitions, layer, *args, **kwargs)` {#repeat} + +Applies the same layer with the same arguments repeatedly. + +```python + y = repeat(x, 3, conv2d, 64, [3, 3], scope='conv1') + # It is equivalent to: + + x = conv2d(x, 64, [3, 3], scope='conv1/conv1_1') + x = conv2d(x, 64, [3, 3], scope='conv1/conv1_2') + y = conv2d(x, 64, [3, 3], scope='conv1/conv1_3') +``` + +If the `scope` argument is not given in `kwargs`, it is set to +`layer.__name__`, or `layer.func.__name__` (for `functools.partial` +objects). If neither `__name__` nor `func.__name__` is available, the +layers are called with `scope='stack'`. + +##### Args: + + +* <b>`inputs`</b>: A `Tensor` suitable for layer. +* <b>`repetitions`</b>: Int, number of repetitions. +* <b>`layer`</b>: A layer with arguments `(inputs, *args, **kwargs)` +* <b>`*args`</b>: Extra args for the layer. +* <b>`**kwargs`</b>: Extra kwargs for the layer. + +##### Returns: + + a tensor result of applying the layer, repetitions times. + +##### Raises: + + +* <b>`ValueError`</b>: if the op is unknown or wrong. + + +- - - + +### `tf.contrib.layers.separable_convolution2d(*args, **kwargs)` {#separable_convolution2d} + +Adds a depth-separable 2D convolution with optional batch_norm layer. + +This op first performs a depthwise convolution that acts separately on +channels, creating a variable called `depthwise_weights`. If `num_outputs` +is not None, it adds a pointwise convolution that mixes channels, creating a +variable called `pointwise_weights`. Then, if `batch_norm_params` is None, +it adds bias to the result, creating a variable called 'biases', otherwise +it adds a batch normalization layer. It finally applies an activation function +to produce the end result. + +##### Args: + + +* <b>`inputs`</b>: a tensor of size [batch_size, height, width, channels]. +* <b>`num_outputs`</b>: the number of pointwise convolution output filters. If is + None, then we skip the pointwise convolution stage. +* <b>`kernel_size`</b>: a list of length 2: [kernel_height, kernel_width] of + of the filters. Can be an int if both values are the same. +* <b>`depth_multiplier`</b>: the number of depthwise convolution output channels for + each input channel. The total number of depthwise convolution output + channels will be equal to `num_filters_in * depth_multiplier`. +* <b>`stride`</b>: a list of length 2: [stride_height, stride_width], specifying the + depthwise convolution stride. Can be an int if both strides are the same. +* <b>`padding`</b>: one of 'VALID' or 'SAME'. +* <b>`activation_fn`</b>: activation function. +* <b>`normalizer_fn`</b>: normalization function to use instead of `biases`. If + `normalize_fn` is provided then `biases_initializer` and + `biases_regularizer` are ignored and `biases` are not created nor added. +* <b>`normalizer_params`</b>: normalization function parameters. +* <b>`weights_initializer`</b>: An initializer for the weights. +* <b>`weights_regularizer`</b>: Optional regularizer for the weights. +* <b>`biases_initializer`</b>: An initializer for the biases. If None skip biases. +* <b>`biases_regularizer`</b>: Optional regularizer for the biases. +* <b>`reuse`</b>: whether or not the layer and its variables should be reused. To be + able to reuse the layer scope must be given. +* <b>`variables_collections`</b>: optional list of collections for all the variables or + a dictionay containing a different list of collection per variable. +* <b>`outputs_collections`</b>: collection to add the outputs. +* <b>`trainable`</b>: whether or not the variables should be trainable or not. +* <b>`scope`</b>: Optional scope for variable_op_scope. + +##### Returns: + + A `Tensor` representing the output of the operation. + + +- - - + +### `tf.contrib.layers.stack(inputs, layer, stack_args, **kwargs)` {#stack} + +Builds a stack of layers by applying layer repeatedly using stack_args. + +`stack` allows you to repeatedly apply the same operation with different +arguments `stack_args[i]`. For each application of the layer, `stack` creates +a new scope appended with an increasing number. For example: + +```python + y = stack(x, fully_connected, [32, 64, 128], scope='fc') + # It is equivalent to: + + x = fully_connected(x, 32, scope='fc/fc_1') + x = fully_connected(x, 64, scope='fc/fc_2') + y = fully_connected(x, 128, scope='fc/fc_3') +``` + +If the `scope` argument is not given in `kwargs`, it is set to +`layer.__name__`, or `layer.func.__name__` (for `functools.partial` +objects). If neither `__name__` nor `func.__name__` is available, the +layers are called with `scope='stack'`. + +##### Args: + + +* <b>`inputs`</b>: A `Tensor` suitable for layer. +* <b>`layer`</b>: A layer with arguments `(inputs, *args, **kwargs)` +* <b>`stack_args`</b>: A list/tuple of parameters for each call of layer. +* <b>`**kwargs`</b>: Extra kwargs for the layer. + +##### Returns: + + a `Tensor` result of applying the stacked layers. + +##### Raises: + + +* <b>`ValueError`</b>: if the op is unknown or wrong. + + +- - - + +### `tf.contrib.layers.unit_norm(*args, **kwargs)` {#unit_norm} + +Normalizes the given input across the specified dimension to unit length. + +Note that the rank of `input` must be known. + +##### Args: + + +* <b>`inputs`</b>: A `Tensor` of arbitrary size. +* <b>`dim`</b>: The dimension along which the input is normalized. +* <b>`epsilon`</b>: A small value to add to the inputs to avoid dividing by zero. +* <b>`scope`</b>: Optional scope for variable_op_scope. + +##### Returns: + + The normalized `Tensor`. + +##### Raises: + + +* <b>`ValueError`</b>: If dim is smaller than the number of dimensions in 'inputs'. + + Aliases for fully_connected which set a default activation function are available: `relu`, `relu6` and `linear`. diff --git a/tensorflow/g3doc/api_docs/python/functions_and_classes/shard0/tf.contrib.layers.convolution2d_in_plane.md b/tensorflow/g3doc/api_docs/python/functions_and_classes/shard0/tf.contrib.layers.convolution2d_in_plane.md new file mode 100644 index 0000000000..2de9c7e8e9 --- /dev/null +++ b/tensorflow/g3doc/api_docs/python/functions_and_classes/shard0/tf.contrib.layers.convolution2d_in_plane.md @@ -0,0 +1,47 @@ +### `tf.contrib.layers.convolution2d_in_plane(*args, **kwargs)` {#convolution2d_in_plane} + +Performs the same in-plane convolution to each channel independently. + +This is useful for performing various simple channel-independent convolution +operations such as image gradients: + + image = tf.constant(..., shape=(16, 240, 320, 3)) + vert_gradients = layers.conv2d_in_plane(image, + kernel=[1, -1], + kernel_size=[2, 1]) + horz_gradients = layers.conv2d_in_plane(image, + kernel=[1, -1], + kernel_size=[1, 2]) + +##### Args: + + +* <b>`inputs`</b>: a 4-D tensor with dimensions [batch_size, height, width, channels]. +* <b>`kernel_size`</b>: a list of length 2 holding the [kernel_height, kernel_width] of + of the pooling. Can be an int if both values are the same. +* <b>`stride`</b>: a list of length 2 `[stride_height, stride_width]`. + Can be an int if both strides are the same. Note that presently + both strides must have the same value. +* <b>`padding`</b>: the padding type to use, either 'SAME' or 'VALID'. +* <b>`activation_fn`</b>: activation function. +* <b>`normalizer_fn`</b>: normalization function to use instead of `biases`. If + `normalize_fn` is provided then `biases_initializer` and + `biases_regularizer` are ignored and `biases` are not created nor added. +* <b>`normalizer_params`</b>: normalization function parameters. +* <b>`weights_initializer`</b>: An initializer for the weights. +* <b>`weights_regularizer`</b>: Optional regularizer for the weights. +* <b>`biases_initializer`</b>: An initializer for the biases. If None skip biases. +* <b>`biases_regularizer`</b>: Optional regularizer for the biases. +* <b>`reuse`</b>: whether or not the layer and its variables should be reused. To be + able to reuse the layer scope must be given. +* <b>`variables_collections`</b>: optional list of collections for all the variables or + a dictionay containing a different list of collection per variable. +* <b>`outputs_collections`</b>: collection to add the outputs. +* <b>`trainable`</b>: If `True` also add variables to the graph collection + `GraphKeys.TRAINABLE_VARIABLES` (see tf.Variable). +* <b>`scope`</b>: Optional scope for `variable_op_scope`. + +##### Returns: + + A `Tensor` representing the output of the operation. + diff --git a/tensorflow/g3doc/api_docs/python/functions_and_classes/shard1/tf.contrib.layers.flatten.md b/tensorflow/g3doc/api_docs/python/functions_and_classes/shard1/tf.contrib.layers.flatten.md new file mode 100644 index 0000000000..29a19d29c6 --- /dev/null +++ b/tensorflow/g3doc/api_docs/python/functions_and_classes/shard1/tf.contrib.layers.flatten.md @@ -0,0 +1,22 @@ +### `tf.contrib.layers.flatten(*args, **kwargs)` {#flatten} + +Flattens the input while maintaining the batch_size. + + Assumes that the first dimension represents the batch. + +##### Args: + + +* <b>`inputs`</b>: a tensor of size [batch_size, ...]. +* <b>`outputs_collections`</b>: collection to add the outputs. +* <b>`scope`</b>: Optional scope for op_scope. + +##### Returns: + + a flattened tensor with shape [batch_size, k]. + +##### Raises: + + +* <b>`ValueError`</b>: if inputs.shape is wrong. + diff --git a/tensorflow/g3doc/api_docs/python/functions_and_classes/shard2/tf.contrib.layers.repeat.md b/tensorflow/g3doc/api_docs/python/functions_and_classes/shard2/tf.contrib.layers.repeat.md new file mode 100644 index 0000000000..24eac3e288 --- /dev/null +++ b/tensorflow/g3doc/api_docs/python/functions_and_classes/shard2/tf.contrib.layers.repeat.md @@ -0,0 +1,36 @@ +### `tf.contrib.layers.repeat(inputs, repetitions, layer, *args, **kwargs)` {#repeat} + +Applies the same layer with the same arguments repeatedly. + +```python + y = repeat(x, 3, conv2d, 64, [3, 3], scope='conv1') + # It is equivalent to: + + x = conv2d(x, 64, [3, 3], scope='conv1/conv1_1') + x = conv2d(x, 64, [3, 3], scope='conv1/conv1_2') + y = conv2d(x, 64, [3, 3], scope='conv1/conv1_3') +``` + +If the `scope` argument is not given in `kwargs`, it is set to +`layer.__name__`, or `layer.func.__name__` (for `functools.partial` +objects). If neither `__name__` nor `func.__name__` is available, the +layers are called with `scope='stack'`. + +##### Args: + + +* <b>`inputs`</b>: A `Tensor` suitable for layer. +* <b>`repetitions`</b>: Int, number of repetitions. +* <b>`layer`</b>: A layer with arguments `(inputs, *args, **kwargs)` +* <b>`*args`</b>: Extra args for the layer. +* <b>`**kwargs`</b>: Extra kwargs for the layer. + +##### Returns: + + a tensor result of applying the layer, repetitions times. + +##### Raises: + + +* <b>`ValueError`</b>: if the op is unknown or wrong. + diff --git a/tensorflow/g3doc/api_docs/python/functions_and_classes/shard3/tf.contrib.layers.separable_convolution2d.md b/tensorflow/g3doc/api_docs/python/functions_and_classes/shard3/tf.contrib.layers.separable_convolution2d.md new file mode 100644 index 0000000000..cd8dc35151 --- /dev/null +++ b/tensorflow/g3doc/api_docs/python/functions_and_classes/shard3/tf.contrib.layers.separable_convolution2d.md @@ -0,0 +1,47 @@ +### `tf.contrib.layers.separable_convolution2d(*args, **kwargs)` {#separable_convolution2d} + +Adds a depth-separable 2D convolution with optional batch_norm layer. + +This op first performs a depthwise convolution that acts separately on +channels, creating a variable called `depthwise_weights`. If `num_outputs` +is not None, it adds a pointwise convolution that mixes channels, creating a +variable called `pointwise_weights`. Then, if `batch_norm_params` is None, +it adds bias to the result, creating a variable called 'biases', otherwise +it adds a batch normalization layer. It finally applies an activation function +to produce the end result. + +##### Args: + + +* <b>`inputs`</b>: a tensor of size [batch_size, height, width, channels]. +* <b>`num_outputs`</b>: the number of pointwise convolution output filters. If is + None, then we skip the pointwise convolution stage. +* <b>`kernel_size`</b>: a list of length 2: [kernel_height, kernel_width] of + of the filters. Can be an int if both values are the same. +* <b>`depth_multiplier`</b>: the number of depthwise convolution output channels for + each input channel. The total number of depthwise convolution output + channels will be equal to `num_filters_in * depth_multiplier`. +* <b>`stride`</b>: a list of length 2: [stride_height, stride_width], specifying the + depthwise convolution stride. Can be an int if both strides are the same. +* <b>`padding`</b>: one of 'VALID' or 'SAME'. +* <b>`activation_fn`</b>: activation function. +* <b>`normalizer_fn`</b>: normalization function to use instead of `biases`. If + `normalize_fn` is provided then `biases_initializer` and + `biases_regularizer` are ignored and `biases` are not created nor added. +* <b>`normalizer_params`</b>: normalization function parameters. +* <b>`weights_initializer`</b>: An initializer for the weights. +* <b>`weights_regularizer`</b>: Optional regularizer for the weights. +* <b>`biases_initializer`</b>: An initializer for the biases. If None skip biases. +* <b>`biases_regularizer`</b>: Optional regularizer for the biases. +* <b>`reuse`</b>: whether or not the layer and its variables should be reused. To be + able to reuse the layer scope must be given. +* <b>`variables_collections`</b>: optional list of collections for all the variables or + a dictionay containing a different list of collection per variable. +* <b>`outputs_collections`</b>: collection to add the outputs. +* <b>`trainable`</b>: whether or not the variables should be trainable or not. +* <b>`scope`</b>: Optional scope for variable_op_scope. + +##### Returns: + + A `Tensor` representing the output of the operation. + diff --git a/tensorflow/g3doc/api_docs/python/functions_and_classes/shard4/tf.contrib.layers.avg_pool2d.md b/tensorflow/g3doc/api_docs/python/functions_and_classes/shard4/tf.contrib.layers.avg_pool2d.md new file mode 100644 index 0000000000..b10aeaef09 --- /dev/null +++ b/tensorflow/g3doc/api_docs/python/functions_and_classes/shard4/tf.contrib.layers.avg_pool2d.md @@ -0,0 +1,25 @@ +### `tf.contrib.layers.avg_pool2d(*args, **kwargs)` {#avg_pool2d} + +Adds a Avg Pooling op. + +It is assumed by the wrapper that the pooling is only done per image and not +in depth or batch. + +##### Args: + + +* <b>`inputs`</b>: a tensor of size [batch_size, height, width, depth]. +* <b>`kernel_size`</b>: a list of length 2: [kernel_height, kernel_width] of the + pooling kernel over which the op is computed. Can be an int if both + values are the same. +* <b>`stride`</b>: a list of length 2: [stride_height, stride_width]. + Can be an int if both strides are the same. Note that presently + both strides must have the same value. +* <b>`padding`</b>: the padding method, either 'VALID' or 'SAME'. +* <b>`outputs_collections`</b>: collection to add the outputs. +* <b>`scope`</b>: Optional scope for op_scope. + +##### Returns: + + a tensor representing the results of the pooling operation. + diff --git a/tensorflow/g3doc/api_docs/python/functions_and_classes/shard4/tf.contrib.layers.batch_norm.md b/tensorflow/g3doc/api_docs/python/functions_and_classes/shard4/tf.contrib.layers.batch_norm.md new file mode 100644 index 0000000000..bc7498e5a5 --- /dev/null +++ b/tensorflow/g3doc/api_docs/python/functions_and_classes/shard4/tf.contrib.layers.batch_norm.md @@ -0,0 +1,48 @@ +### `tf.contrib.layers.batch_norm(*args, **kwargs)` {#batch_norm} + +Adds a Batch Normalization layer from http://arxiv.org/abs/1502.03167. + + "Batch Normalization: Accelerating Deep Network Training by Reducing + Internal Covariate Shift" + + Sergey Ioffe, Christian Szegedy + +Can be used as a normalizer function for conv2d and fully_connected. + +##### Args: + + +* <b>`inputs`</b>: a tensor of size `[batch_size, height, width, channels]` + or `[batch_size, channels]`. +* <b>`decay`</b>: decay for the moving average. +* <b>`center`</b>: If True, subtract `beta`. If False, `beta` is ignored. +* <b>`scale`</b>: If True, multiply by `gamma`. If False, `gamma` is + not used. When the next layer is linear (also e.g. `nn.relu`), this can be + disabled since the scaling can be done by the next layer. +* <b>`epsilon`</b>: small float added to variance to avoid dividing by zero. +* <b>`activation_fn`</b>: Optional activation function. +* <b>`updates_collections`</b>: collections to collect the update ops for computation. + If None, a control dependency would be added to make sure the updates are + computed. +* <b>`is_training`</b>: whether or not the layer is in training mode. In training mode + it would accumulate the statistics of the moments into `moving_mean` and + `moving_variance` using an exponential moving average with the given + `decay`. When it is not in training mode then it would use the values of + the `moving_mean` and the `moving_variance`. +* <b>`reuse`</b>: whether or not the layer and its variables should be reused. To be + able to reuse the layer scope must be given. +* <b>`variables_collections`</b>: optional collections for the variables. +* <b>`outputs_collections`</b>: collections to add the outputs. +* <b>`trainable`</b>: If `True` also add variables to the graph collection + `GraphKeys.TRAINABLE_VARIABLES` (see tf.Variable). +* <b>`scope`</b>: Optional scope for `variable_op_scope`. + +##### Returns: + + A `Tensor` representing the output of the operation. + +##### Raises: + + +* <b>`ValueError`</b>: if rank or last dimension of `inputs` is undefined. + diff --git a/tensorflow/g3doc/api_docs/python/functions_and_classes/shard4/tf.contrib.layers.unit_norm.md b/tensorflow/g3doc/api_docs/python/functions_and_classes/shard4/tf.contrib.layers.unit_norm.md new file mode 100644 index 0000000000..ee1954ffc0 --- /dev/null +++ b/tensorflow/g3doc/api_docs/python/functions_and_classes/shard4/tf.contrib.layers.unit_norm.md @@ -0,0 +1,23 @@ +### `tf.contrib.layers.unit_norm(*args, **kwargs)` {#unit_norm} + +Normalizes the given input across the specified dimension to unit length. + +Note that the rank of `input` must be known. + +##### Args: + + +* <b>`inputs`</b>: A `Tensor` of arbitrary size. +* <b>`dim`</b>: The dimension along which the input is normalized. +* <b>`epsilon`</b>: A small value to add to the inputs to avoid dividing by zero. +* <b>`scope`</b>: Optional scope for variable_op_scope. + +##### Returns: + + The normalized `Tensor`. + +##### Raises: + + +* <b>`ValueError`</b>: If dim is smaller than the number of dimensions in 'inputs'. + diff --git a/tensorflow/g3doc/api_docs/python/functions_and_classes/shard5/tf.contrib.layers.one_hot_encoding.md b/tensorflow/g3doc/api_docs/python/functions_and_classes/shard5/tf.contrib.layers.one_hot_encoding.md new file mode 100644 index 0000000000..0b0d3d8e9a --- /dev/null +++ b/tensorflow/g3doc/api_docs/python/functions_and_classes/shard5/tf.contrib.layers.one_hot_encoding.md @@ -0,0 +1,18 @@ +### `tf.contrib.layers.one_hot_encoding(*args, **kwargs)` {#one_hot_encoding} + +Transform numeric labels into onehot_labels using tf.one_hot. + +##### Args: + + +* <b>`labels`</b>: [batch_size] target labels. +* <b>`num_classes`</b>: total number of classes. +* <b>`on_value`</b>: A scalar defining the on-value. +* <b>`off_value`</b>: A scalar defining the off-value. +* <b>`outputs_collections`</b>: collection to add the outputs. +* <b>`scope`</b>: Optional scope for op_scope. + +##### Returns: + + one hot encoding of the labels. + diff --git a/tensorflow/g3doc/api_docs/python/functions_and_classes/shard5/tf.contrib.layers.stack.md b/tensorflow/g3doc/api_docs/python/functions_and_classes/shard5/tf.contrib.layers.stack.md new file mode 100644 index 0000000000..f387553830 --- /dev/null +++ b/tensorflow/g3doc/api_docs/python/functions_and_classes/shard5/tf.contrib.layers.stack.md @@ -0,0 +1,39 @@ +### `tf.contrib.layers.stack(inputs, layer, stack_args, **kwargs)` {#stack} + +Builds a stack of layers by applying layer repeatedly using stack_args. + +`stack` allows you to repeatedly apply the same operation with different +arguments `stack_args[i]`. For each application of the layer, `stack` creates +a new scope appended with an increasing number. For example: + +```python + y = stack(x, fully_connected, [32, 64, 128], scope='fc') + # It is equivalent to: + + x = fully_connected(x, 32, scope='fc/fc_1') + x = fully_connected(x, 64, scope='fc/fc_2') + y = fully_connected(x, 128, scope='fc/fc_3') +``` + +If the `scope` argument is not given in `kwargs`, it is set to +`layer.__name__`, or `layer.func.__name__` (for `functools.partial` +objects). If neither `__name__` nor `func.__name__` is available, the +layers are called with `scope='stack'`. + +##### Args: + + +* <b>`inputs`</b>: A `Tensor` suitable for layer. +* <b>`layer`</b>: A layer with arguments `(inputs, *args, **kwargs)` +* <b>`stack_args`</b>: A list/tuple of parameters for each call of layer. +* <b>`**kwargs`</b>: Extra kwargs for the layer. + +##### Returns: + + a `Tensor` result of applying the stacked layers. + +##### Raises: + + +* <b>`ValueError`</b>: if the op is unknown or wrong. + diff --git a/tensorflow/g3doc/api_docs/python/functions_and_classes/shard7/tf.contrib.layers.convolution2d_transpose.md b/tensorflow/g3doc/api_docs/python/functions_and_classes/shard7/tf.contrib.layers.convolution2d_transpose.md new file mode 100644 index 0000000000..9251a30908 --- /dev/null +++ b/tensorflow/g3doc/api_docs/python/functions_and_classes/shard7/tf.contrib.layers.convolution2d_transpose.md @@ -0,0 +1,45 @@ +### `tf.contrib.layers.convolution2d_transpose(*args, **kwargs)` {#convolution2d_transpose} + +Adds a convolution2d_transpose with an optional batch normalization layer. + +The function creates a variable called `weights`, representing the +kernel, that is convolved with the input. If `batch_norm_params` is `None`, a +second variable called 'biases' is added to the result of the operation. + +##### Args: + + +* <b>`inputs`</b>: a tensor of size [batch_size, height, width, channels]. +* <b>`num_outputs`</b>: integer, the number of output filters. +* <b>`kernel_size`</b>: a list of length 2 holding the [kernel_height, kernel_width] of + of the filters. Can be an int if both values are the same. +* <b>`stride`</b>: a list of length 2: [stride_height, stride_width]. + Can be an int if both strides are the same. Note that presently + both strides must have the same value. +* <b>`padding`</b>: one of 'VALID' or 'SAME'. +* <b>`activation_fn`</b>: activation function. +* <b>`normalizer_fn`</b>: normalization function to use instead of `biases`. If + `normalize_fn` is provided then `biases_initializer` and + `biases_regularizer` are ignored and `biases` are not created nor added. +* <b>`normalizer_params`</b>: normalization function parameters. +* <b>`weights_initializer`</b>: An initializer for the weights. +* <b>`weights_regularizer`</b>: Optional regularizer for the weights. +* <b>`biases_initializer`</b>: An initializer for the biases. If None skip biases. +* <b>`biases_regularizer`</b>: Optional regularizer for the biases. +* <b>`reuse`</b>: whether or not the layer and its variables should be reused. To be + able to reuse the layer scope must be given. +* <b>`variables_collections`</b>: optional list of collections for all the variables or + a dictionay containing a different list of collection per variable. +* <b>`outputs_collections`</b>: collection to add the outputs. +* <b>`trainable`</b>: whether or not the variables should be trainable or not. +* <b>`scope`</b>: Optional scope for variable_op_scope. + +##### Returns: + + a tensor representing the output of the operation. + +##### Raises: + + +* <b>`ValueError`</b>: if 'kernel_size' is not a list of length 2. + diff --git a/tensorflow/g3doc/api_docs/python/functions_and_classes/shard8/tf.contrib.layers.max_pool2d.md b/tensorflow/g3doc/api_docs/python/functions_and_classes/shard8/tf.contrib.layers.max_pool2d.md new file mode 100644 index 0000000000..5dd8fbf68d --- /dev/null +++ b/tensorflow/g3doc/api_docs/python/functions_and_classes/shard8/tf.contrib.layers.max_pool2d.md @@ -0,0 +1,30 @@ +### `tf.contrib.layers.max_pool2d(*args, **kwargs)` {#max_pool2d} + +Adds a Max Pooling op. + +It is assumed by the wrapper that the pooling is only done per image and not +in depth or batch. + +##### Args: + + +* <b>`inputs`</b>: a tensor of size [batch_size, height, width, depth]. +* <b>`kernel_size`</b>: a list of length 2: [kernel_height, kernel_width] of the + pooling kernel over which the op is computed. Can be an int if both + values are the same. +* <b>`stride`</b>: a list of length 2: [stride_height, stride_width]. + Can be an int if both strides are the same. Note that presently + both strides must have the same value. +* <b>`padding`</b>: the padding method, either 'VALID' or 'SAME'. +* <b>`outputs_collections`</b>: collection to add the outputs. +* <b>`scope`</b>: Optional scope for op_scope. + +##### Returns: + + a tensor representing the results of the pooling operation. + +##### Raises: + + +* <b>`ValueError`</b>: if 'kernel_size' is not a 2-D list + diff --git a/tensorflow/g3doc/api_docs/python/index.md b/tensorflow/g3doc/api_docs/python/index.md index e6c568ec0d..aed0c56eb0 100644 --- a/tensorflow/g3doc/api_docs/python/index.md +++ b/tensorflow/g3doc/api_docs/python/index.md @@ -657,17 +657,28 @@ * **[Layers (contrib)](../../api_docs/python/contrib.layers.md)**: * [`apply_regularization`](../../api_docs/python/contrib.layers.md#apply_regularization) + * [`avg_pool2d`](../../api_docs/python/contrib.layers.md#avg_pool2d) + * [`batch_norm`](../../api_docs/python/contrib.layers.md#batch_norm) * [`convolution2d`](../../api_docs/python/contrib.layers.md#convolution2d) + * [`convolution2d_in_plane`](../../api_docs/python/contrib.layers.md#convolution2d_in_plane) + * [`convolution2d_transpose`](../../api_docs/python/contrib.layers.md#convolution2d_transpose) + * [`flatten`](../../api_docs/python/contrib.layers.md#flatten) * [`fully_connected`](../../api_docs/python/contrib.layers.md#fully_connected) * [`l1_regularizer`](../../api_docs/python/contrib.layers.md#l1_regularizer) * [`l2_regularizer`](../../api_docs/python/contrib.layers.md#l2_regularizer) + * [`max_pool2d`](../../api_docs/python/contrib.layers.md#max_pool2d) + * [`one_hot_encoding`](../../api_docs/python/contrib.layers.md#one_hot_encoding) * [`optimize_loss`](../../api_docs/python/contrib.layers.md#optimize_loss) + * [`repeat`](../../api_docs/python/contrib.layers.md#repeat) + * [`separable_convolution2d`](../../api_docs/python/contrib.layers.md#separable_convolution2d) + * [`stack`](../../api_docs/python/contrib.layers.md#stack) * [`sum_regularizer`](../../api_docs/python/contrib.layers.md#sum_regularizer) * [`summarize_activation`](../../api_docs/python/contrib.layers.md#summarize_activation) * [`summarize_activations`](../../api_docs/python/contrib.layers.md#summarize_activations) * [`summarize_collection`](../../api_docs/python/contrib.layers.md#summarize_collection) * [`summarize_tensor`](../../api_docs/python/contrib.layers.md#summarize_tensor) * [`summarize_tensors`](../../api_docs/python/contrib.layers.md#summarize_tensors) + * [`unit_norm`](../../api_docs/python/contrib.layers.md#unit_norm) * [`variance_scaling_initializer`](../../api_docs/python/contrib.layers.md#variance_scaling_initializer) * [`xavier_initializer`](../../api_docs/python/contrib.layers.md#xavier_initializer) * [`xavier_initializer_conv2d`](../../api_docs/python/contrib.layers.md#xavier_initializer_conv2d) |