B ӻdF8 @sdZddlZddlZddlmZddlmZddlm Z ddl m Z ddl m Z ddl mZdd l mZdd l mZdd l mZdd lmZdd lmZddlmZddlmZddlmZddlmZddlmZddlmZddlmZddlm Z ddl!m"Z"ddl!m#Z#ddl!m$Z$ddl%m&Z&ddl'm(Z(ddl)m*Z*e(dGdddZ+Gdd d e+Z,e(d!Gd"d#d#e,Z-e(d$Gd%d&d&e,Z.e(d'Gd(d)d)e,Z/e(d*Gd+d,d,e,Z0e(d-Gd.d/d/e,Z1e(d0Gd1d2d2e,Z2e(d3Gd4d5d5e,Z3e(d6Gd7d8d8e,Z4e(d9Gd:d;d;e,Z5e(d<Gd=d>d>e,Z6e(d?Gd@dAdAe,Z7e(dBGdCdDdDe,Z8e(dEGdFdGdGe,Z9e(dHGdIdJdJe,Z:e(dKdLdMdNdOdPe&j;dQdRZeeAe#j?d`daZBe(dbdcdddedfdge&j;dhdiZCe&>eCe#j?djdkZDe(dldmdndodpdqe&j;drdsZEe&>eEe#j?dtduZFdvdwZGe(dxdye&j;dzd{ZHe(d|d}e&j;d~dZIe(de&j;ddZJe(dgde&j;dddZKe(dddde&j;ddZLe(dde&j;dddZMe&>eMe#j?dddZNe(dde&j;dddZOe&>eOe#j?dddZPe(dde&j;dddZQe&>eQe#j?dddZRe(dddddddde&j;ddZSe(dde&j;ddZTe(ddddddgde&j;dddZUe(dGddde,ZVeQZWZXe<ZYZZeAZ[Z\eCZ]Z^eEZ_Z`eSZaZbZceLZdeKZeddZfe(dddZge(ddddZhe(dddÄZie jjdeOdiZkdS)zBuilt-in loss functions.N)ag_ctx)api)distribution_strategy_context)context) constant_op)ops) smart_cond) tensor_spec) tensor_util)backend) losses_utils)tf_utils)deserialize_keras_object)serialize_keras_object) array_ops)control_flow_ops)math_ops)nn) losses_impl)ragged_map_ops) ragged_tensor) ragged_util)dispatch) keras_export) doc_controlszkeras.losses.Lossc@sdeZdZdZejjdfddZddZdddZ e d d Z d d Z e jejd dZddZdS)LossaLoss base class. To be implemented by subclasses: * `call()`: Contains the logic for loss calculation using `y_true`, `y_pred`. Example subclass implementation: ```python class MeanSquaredError(Loss): def call(self, y_true, y_pred): y_pred = tf.convert_to_tensor_v2(y_pred) y_true = tf.cast(y_true, y_pred.dtype) return tf.reduce_mean(math_ops.square(y_pred - y_true), axis=-1) ``` When used with `tf.distribute.Strategy`, outside of built-in training loops such as `tf.keras` `compile` and `fit`, please use 'SUM' or 'NONE' reduction types, and reduce losses explicitly in your training loop. Using 'AUTO' or 'SUM_OVER_BATCH_SIZE' will raise an error. Please see this custom training [tutorial]( https://www.tensorflow.org/tutorials/distribute/custom_training) for more details on this. You can implement 'SUM_OVER_BATCH_SIZE' using global batch size like: ```python with strategy.scope(): loss_obj = tf.keras.losses.CategoricalCrossentropy( reduction=tf.keras.losses.Reduction.NONE) .... loss = (tf.reduce_sum(loss_obj(labels, predictions)) * (1. / global_batch_size)) ``` NcCs*tj|||_||_d|_|dS)aInitializes `Loss` class. Args: reduction: Type of `tf.keras.losses.Reduction` to apply to loss. Default value is `AUTO`. `AUTO` indicates that the reduction option will be determined by the usage context. For almost all cases this defaults to `SUM_OVER_BATCH_SIZE`. When used with `tf.distribute.Strategy`, outside of built-in training loops such as `tf.keras` `compile` and `fit`, using `AUTO` or `SUM_OVER_BATCH_SIZE` will raise an error. Please see this custom training [tutorial]( https://www.tensorflow.org/tutorials/distribute/custom_training) for more details. name: Optional name for the instance. FN)r ReductionV2validate reductionname_allow_sum_over_batch_size_set_name_scope)selfrrr#P/var/www/html/venv/lib/python3.7/site-packages/tensorflow/python/keras/losses.py__init__Ws  z Loss.__init__cCs:|jdkr|jj|_n |jdkr(d|_n|jd|_dS)z"Creates a valid `name_scope` name.Nzlambda_)r __class____name__ _name_scopestrip)r"r#r#r$r!ns    zLoss._set_name_scopec Csvt|||}t|jR|Btr2|j}nt |jt }|||}t j |||dSQRXWdQRXdS)aInvokes the `Loss` instance. Args: y_true: Ground truth values. shape = `[batch_size, d0, .. dN]`, except sparse loss functions such as sparse categorical crossentropy where shape = `[batch_size, d0, .. dN-1]` y_pred: The predicted values. shape = `[batch_size, d0, .. dN]` sample_weight: Optional `sample_weight` acts as a coefficient for the loss. If a scalar is provided, then the loss is simply scaled by the given value. If `sample_weight` is a tensor of size `[batch_size]`, then the total loss for each sample of the batch is rescaled by the corresponding element in the `sample_weight` vector. If the shape of `sample_weight` is `[batch_size, d0, .. dN-1]` (or can be broadcasted to this shape), then each loss element of `y_pred` is scaled by the corresponding value of `sample_weight`. (Note on`dN-1`: all loss functions reduce by 1 dimension, usually axis=-1.) Returns: Weighted loss float `Tensor`. If `reduction` is `NONE`, this has shape `[batch_size, d0, .. dN-1]`; otherwise, it is scalar. (Note `dN-1` because all loss functions reduce by 1 dimension, usually axis=-1.) Raises: ValueError: If the shape of `sample_weight` is invalid. )rN)r Z"graph_context_for_symbolic_tensorsr Z name_scoper*rZexecuting_eagerlycall autograph tf_convertrcontrol_status_ctxr Zcompute_weighted_loss_get_reduction)r"y_truey_predZ sample_weightZ graph_ctxZcall_fnZlossesr#r#r$__call__xs  z Loss.__call__cCs |f|S)zInstantiates a `Loss` from its config (output of `get_config()`). Args: config: Output of `get_config()`. Returns: A `Loss` instance. r#)clsconfigr#r#r$ from_configs zLoss.from_configcCs|j|jdS)z4Returns the config dictionary for a `Loss` instance.)rr)rr)r"r#r#r$ get_configszLoss.get_configcCs tddS)aInvokes the `Loss` instance. Args: y_true: Ground truth values. shape = `[batch_size, d0, .. dN]`, except sparse loss functions such as sparse categorical crossentropy where shape = `[batch_size, d0, .. dN-1]` y_pred: The predicted values. shape = `[batch_size, d0, .. dN]` Returns: Loss values with the shape `[batch_size, d0, .. dN-1]`. z"Must be implemented in subclasses.N)NotImplementedError)r"r1r2r#r#r$r,sz Loss.callcCsN|js2tr2|jtjjks*|jtjjkr2td|jtjjkrHtjjS|jS)z?Handles `AUTO` reduction cases and returns the reduction value.aQPlease use `tf.keras.losses.Reduction.SUM` or `tf.keras.losses.Reduction.NONE` for loss reduction when losses are used with `tf.distribute.Strategy` outside of the built-in training loops. You can implement `tf.keras.losses.Reduction.SUM_OVER_BATCH_SIZE` using global batch size like: ``` with strategy.scope(): loss_obj = tf.keras.losses.CategoricalCrossentropy(reduction=tf.keras.losses.Reduction.NONE) .... loss = tf.reduce_sum(loss_obj(labels, predictions)) * (1. / global_batch_size) ``` Please see https://www.tensorflow.org/tutorials/distribute/custom_training for more details.) r rZ has_strategyrr rAUTOZSUM_OVER_BATCH_SIZE ValueError)r"r#r#r$r0s zLoss._get_reduction)N)r) __module__ __qualname____doc__r rr9r%r!r3 classmethodr6r7abcabstractmethodrZfor_subclass_implementersr,r0r#r#r#r$r0s% ' rcs>eZdZdZejjdffdd ZddZfddZ Z S) LossFunctionWrapperz*Wraps a loss function in the `Loss` class.Nc s tj||d||_||_dS)agInitializes `LossFunctionWrapper` class. Args: fn: The loss function to wrap, with signature `fn(y_true, y_pred, **kwargs)`. reduction: Type of `tf.keras.losses.Reduction` to apply to loss. Default value is `AUTO`. `AUTO` indicates that the reduction option will be determined by the usage context. For almost all cases this defaults to `SUM_OVER_BATCH_SIZE`. When used with `tf.distribute.Strategy`, outside of built-in training loops such as `tf.keras` `compile` and `fit`, using `AUTO` or `SUM_OVER_BATCH_SIZE` will raise an error. Please see this custom training [tutorial]( https://www.tensorflow.org/tutorials/distribute/custom_training) for more details. name: Optional name for the instance. **kwargs: The keyword arguments that are passed on to `fn`. )rrN)superr%fn _fn_kwargs)r"rCrrkwargs)r(r#r$r%szLossFunctionWrapper.__init__cCsFt|r$t|r$t||\}}t|jt}|||f|j S)zInvokes the `LossFunctionWrapper` instance. Args: y_true: Ground truth values. y_pred: The predicted values. Returns: Loss values per sample. ) r Z is_tf_typer Zsqueeze_or_expand_dimensionsr-r.rCrr/rD)r"r1r2Zag_fnr#r#r$r,s zLossFunctionWrapper.callcs^i}x2|jD]$\}}t|r,t|n|||<qWt}tt |t |S)N) rDitemsr Zis_tensor_or_variabler evalrBr7dictlist)r"r5kvZ base_config)r(r#r$r7s   zLossFunctionWrapper.get_config) r)r;r<r=r rr9r%r,r7 __classcell__r#r#)r(r$rAs rAzkeras.losses.MeanSquaredErrorcs*eZdZdZejjdffdd ZZS)MeanSquaredErroraComputes the mean of squares of errors between labels and predictions. `loss = square(y_true - y_pred)` Standalone usage: >>> y_true = [[0., 1.], [0., 0.]] >>> y_pred = [[1., 1.], [1., 0.]] >>> # Using 'auto'/'sum_over_batch_size' reduction type. >>> mse = tf.keras.losses.MeanSquaredError() >>> mse(y_true, y_pred).numpy() 0.5 >>> # Calling with 'sample_weight'. >>> mse(y_true, y_pred, sample_weight=[0.7, 0.3]).numpy() 0.25 >>> # Using 'sum' reduction type. >>> mse = tf.keras.losses.MeanSquaredError( ... reduction=tf.keras.losses.Reduction.SUM) >>> mse(y_true, y_pred).numpy() 1.0 >>> # Using 'none' reduction type. >>> mse = tf.keras.losses.MeanSquaredError( ... reduction=tf.keras.losses.Reduction.NONE) >>> mse(y_true, y_pred).numpy() array([0.5, 0.5], dtype=float32) Usage with the `compile()` API: ```python model.compile(optimizer='sgd', loss=tf.keras.losses.MeanSquaredError()) ``` mean_squared_errorcstjt||ddS)aInitializes `MeanSquaredError` instance. Args: reduction: Type of `tf.keras.losses.Reduction` to apply to loss. Default value is `AUTO`. `AUTO` indicates that the reduction option will be determined by the usage context. For almost all cases this defaults to `SUM_OVER_BATCH_SIZE`. When used with `tf.distribute.Strategy`, outside of built-in training loops such as `tf.keras` `compile` and `fit`, using `AUTO` or `SUM_OVER_BATCH_SIZE` will raise an error. Please see this custom training [tutorial]( https://www.tensorflow.org/tutorials/distribute/custom_training) for more details. name: Optional name for the instance. Defaults to 'mean_squared_error'. )rrN)rBr%rN)r"rr)r(r#r$r%3szMeanSquaredError.__init__) r)r;r<r=r rr9r%rLr#r#)r(r$rM s$rMzkeras.losses.MeanAbsoluteErrorcs*eZdZdZejjdffdd ZZS)MeanAbsoluteErroraComputes the mean of absolute difference between labels and predictions. `loss = abs(y_true - y_pred)` Standalone usage: >>> y_true = [[0., 1.], [0., 0.]] >>> y_pred = [[1., 1.], [1., 0.]] >>> # Using 'auto'/'sum_over_batch_size' reduction type. >>> mae = tf.keras.losses.MeanAbsoluteError() >>> mae(y_true, y_pred).numpy() 0.5 >>> # Calling with 'sample_weight'. >>> mae(y_true, y_pred, sample_weight=[0.7, 0.3]).numpy() 0.25 >>> # Using 'sum' reduction type. >>> mae = tf.keras.losses.MeanAbsoluteError( ... reduction=tf.keras.losses.Reduction.SUM) >>> mae(y_true, y_pred).numpy() 1.0 >>> # Using 'none' reduction type. >>> mae = tf.keras.losses.MeanAbsoluteError( ... reduction=tf.keras.losses.Reduction.NONE) >>> mae(y_true, y_pred).numpy() array([0.5, 0.5], dtype=float32) Usage with the `compile()` API: ```python model.compile(optimizer='sgd', loss=tf.keras.losses.MeanAbsoluteError()) ``` mean_absolute_errorcstjt||ddS)aInitializes `MeanAbsoluteError` instance. Args: reduction: Type of `tf.keras.losses.Reduction` to apply to loss. Default value is `AUTO`. `AUTO` indicates that the reduction option will be determined by the usage context. For almost all cases this defaults to `SUM_OVER_BATCH_SIZE`. When used with `tf.distribute.Strategy`, outside of built-in training loops such as `tf.keras` `compile` and `fit`, using `AUTO` or `SUM_OVER_BATCH_SIZE` will raise an error. Please see this custom training [tutorial]( https://www.tensorflow.org/tutorials/distribute/custom_training) for more details. name: Optional name for the instance. Defaults to 'mean_absolute_error'. )rrN)rBr%rP)r"rr)r(r#r$r%mszMeanAbsoluteError.__init__) r)r;r<r=r rr9r%rLr#r#)r(r$rOGs$rOz(keras.losses.MeanAbsolutePercentageErrorcs*eZdZdZejjdffdd ZZS)MeanAbsolutePercentageErroraComputes the mean absolute percentage error between `y_true` and `y_pred`. `loss = 100 * abs(y_true - y_pred) / y_true` Standalone usage: >>> y_true = [[2., 1.], [2., 3.]] >>> y_pred = [[1., 1.], [1., 0.]] >>> # Using 'auto'/'sum_over_batch_size' reduction type. >>> mape = tf.keras.losses.MeanAbsolutePercentageError() >>> mape(y_true, y_pred).numpy() 50. >>> # Calling with 'sample_weight'. >>> mape(y_true, y_pred, sample_weight=[0.7, 0.3]).numpy() 20. >>> # Using 'sum' reduction type. >>> mape = tf.keras.losses.MeanAbsolutePercentageError( ... reduction=tf.keras.losses.Reduction.SUM) >>> mape(y_true, y_pred).numpy() 100. >>> # Using 'none' reduction type. >>> mape = tf.keras.losses.MeanAbsolutePercentageError( ... reduction=tf.keras.losses.Reduction.NONE) >>> mape(y_true, y_pred).numpy() array([25., 75.], dtype=float32) Usage with the `compile()` API: ```python model.compile(optimizer='sgd', loss=tf.keras.losses.MeanAbsolutePercentageError()) ``` mean_absolute_percentage_errorcstjt||ddS)a Initializes `MeanAbsolutePercentageError` instance. Args: reduction: Type of `tf.keras.losses.Reduction` to apply to loss. Default value is `AUTO`. `AUTO` indicates that the reduction option will be determined by the usage context. For almost all cases this defaults to `SUM_OVER_BATCH_SIZE`. When used with `tf.distribute.Strategy`, outside of built-in training loops such as `tf.keras` `compile` and `fit`, using `AUTO` or `SUM_OVER_BATCH_SIZE` will raise an error. Please see this custom training [tutorial]( https://www.tensorflow.org/tutorials/distribute/custom_training) for more details. name: Optional name for the instance. Defaults to 'mean_absolute_percentage_error'. )rrN)rBr%rR)r"rr)r(r#r$r%sz$MeanAbsolutePercentageError.__init__) r)r;r<r=r rr9r%rLr#r#)r(r$rQs%rQz(keras.losses.MeanSquaredLogarithmicErrorcs*eZdZdZejjdffdd ZZS)MeanSquaredLogarithmicErrora&Computes the mean squared logarithmic error between `y_true` and `y_pred`. `loss = square(log(y_true + 1.) - log(y_pred + 1.))` Standalone usage: >>> y_true = [[0., 1.], [0., 0.]] >>> y_pred = [[1., 1.], [1., 0.]] >>> # Using 'auto'/'sum_over_batch_size' reduction type. >>> msle = tf.keras.losses.MeanSquaredLogarithmicError() >>> msle(y_true, y_pred).numpy() 0.240 >>> # Calling with 'sample_weight'. >>> msle(y_true, y_pred, sample_weight=[0.7, 0.3]).numpy() 0.120 >>> # Using 'sum' reduction type. >>> msle = tf.keras.losses.MeanSquaredLogarithmicError( ... reduction=tf.keras.losses.Reduction.SUM) >>> msle(y_true, y_pred).numpy() 0.480 >>> # Using 'none' reduction type. >>> msle = tf.keras.losses.MeanSquaredLogarithmicError( ... reduction=tf.keras.losses.Reduction.NONE) >>> msle(y_true, y_pred).numpy() array([0.240, 0.240], dtype=float32) Usage with the `compile()` API: ```python model.compile(optimizer='sgd', loss=tf.keras.losses.MeanSquaredLogarithmicError()) ``` mean_squared_logarithmic_errorcstjt||ddS)a Initializes `MeanSquaredLogarithmicError` instance. Args: reduction: Type of `tf.keras.losses.Reduction` to apply to loss. Default value is `AUTO`. `AUTO` indicates that the reduction option will be determined by the usage context. For almost all cases this defaults to `SUM_OVER_BATCH_SIZE`. When used with `tf.distribute.Strategy`, outside of built-in training loops such as `tf.keras` `compile` and `fit`, using `AUTO` or `SUM_OVER_BATCH_SIZE` will raise an error. Please see this custom training [tutorial]( https://www.tensorflow.org/tutorials/distribute/custom_training) for more details. name: Optional name for the instance. Defaults to 'mean_squared_logarithmic_error'. )rrN)rBr%rT)r"rr)r(r#r$r%sz$MeanSquaredLogarithmicError.__init__) r)r;r<r=r rr9r%rLr#r#)r(r$rSs%rSzkeras.losses.BinaryCrossentropycs0eZdZdZdddejjdffdd ZZS)BinaryCrossentropyaComputes the cross-entropy loss between true labels and predicted labels. Use this cross-entropy loss for binary (0 or 1) classification applications. The loss function requires the following inputs: - `y_true` (true label): This is either 0 or 1. - `y_pred` (predicted value): This is the model's prediction, i.e, a single floating-point value which either represents a [logit](https://en.wikipedia.org/wiki/Logit), (i.e, value in [-inf, inf] when `from_logits=True`) or a probability (i.e, value in [0., 1.] when `from_logits=False`). **Recommended Usage:** (set `from_logits=True`) With `tf.keras` API: ```python model.compile( loss=tf.keras.losses.BinaryCrossentropy(from_logits=True), .... ) ``` As a standalone function: >>> # Example 1: (batch_size = 1, number of samples = 4) >>> y_true = [0, 1, 0, 0] >>> y_pred = [-18.6, 0.51, 2.94, -12.8] >>> bce = tf.keras.losses.BinaryCrossentropy(from_logits=True) >>> bce(y_true, y_pred).numpy() 0.865 >>> # Example 2: (batch_size = 2, number of samples = 4) >>> y_true = [[0, 1], [0, 0]] >>> y_pred = [[-18.6, 0.51], [2.94, -12.8]] >>> # Using default 'auto'/'sum_over_batch_size' reduction type. >>> bce = tf.keras.losses.BinaryCrossentropy(from_logits=True) >>> bce(y_true, y_pred).numpy() 0.865 >>> # Using 'sample_weight' attribute >>> bce(y_true, y_pred, sample_weight=[0.8, 0.2]).numpy() 0.243 >>> # Using 'sum' reduction` type. >>> bce = tf.keras.losses.BinaryCrossentropy(from_logits=True, ... reduction=tf.keras.losses.Reduction.SUM) >>> bce(y_true, y_pred).numpy() 1.730 >>> # Using 'none' reduction type. >>> bce = tf.keras.losses.BinaryCrossentropy(from_logits=True, ... reduction=tf.keras.losses.Reduction.NONE) >>> bce(y_true, y_pred).numpy() array([0.235, 1.496], dtype=float32) **Default Usage:** (set `from_logits=False`) >>> # Make the following updates to the above "Recommended Usage" section >>> # 1. Set `from_logits=False` >>> tf.keras.losses.BinaryCrossentropy() # OR ...('from_logits=False') >>> # 2. Update `y_pred` to use probabilities instead of logits >>> y_pred = [0.6, 0.3, 0.2, 0.8] # OR [[0.6, 0.3], [0.2, 0.8]] Frbinary_crossentropycs"tjt|||||d||_dS)a]Initializes `BinaryCrossentropy` instance. Args: from_logits: Whether to interpret `y_pred` as a tensor of [logit](https://en.wikipedia.org/wiki/Logit) values. By default, we assume that `y_pred` contains probabilities (i.e., values in [0, 1]). label_smoothing: Float in [0, 1]. When 0, no smoothing occurs. When > 0, we compute the loss between the predicted labels and a smoothed version of the true labels, where the smoothing squeezes the labels towards 0.5. Larger values of `label_smoothing` correspond to heavier smoothing. axis: The axis along which to compute crossentropy (the features axis). Defaults to -1. reduction: Type of `tf.keras.losses.Reduction` to apply to loss. Default value is `AUTO`. `AUTO` indicates that the reduction option will be determined by the usage context. For almost all cases this defaults to `SUM_OVER_BATCH_SIZE`. When used with `tf.distribute.Strategy`, outside of built-in training loops such as `tf.keras` `compile` and `fit`, using `AUTO` or `SUM_OVER_BATCH_SIZE` will raise an error. Please see this custom training [tutorial]( https://www.tensorflow.org/tutorials/distribute/custom_training) for more details. name: Name for the op. Defaults to 'binary_crossentropy'. )rr from_logitslabel_smoothingaxisN)rBr%rWrX)r"rXrYrZrr)r(r#r$r%;szBinaryCrossentropy.__init__) r)r;r<r=r rr9r%rLr#r#)r(r$rUs >rUz$keras.losses.CategoricalCrossentropycs0eZdZdZdddejjdffdd ZZS)CategoricalCrossentropyaComputes the crossentropy loss between the labels and predictions. Use this crossentropy loss function when there are two or more label classes. We expect labels to be provided in a `one_hot` representation. If you want to provide labels as integers, please use `SparseCategoricalCrossentropy` loss. There should be `# classes` floating point values per feature. In the snippet below, there is `# classes` floating pointing values per example. The shape of both `y_pred` and `y_true` are `[batch_size, num_classes]`. Standalone usage: >>> y_true = [[0, 1, 0], [0, 0, 1]] >>> y_pred = [[0.05, 0.95, 0], [0.1, 0.8, 0.1]] >>> # Using 'auto'/'sum_over_batch_size' reduction type. >>> cce = tf.keras.losses.CategoricalCrossentropy() >>> cce(y_true, y_pred).numpy() 1.177 >>> # Calling with 'sample_weight'. >>> cce(y_true, y_pred, sample_weight=tf.constant([0.3, 0.7])).numpy() 0.814 >>> # Using 'sum' reduction type. >>> cce = tf.keras.losses.CategoricalCrossentropy( ... reduction=tf.keras.losses.Reduction.SUM) >>> cce(y_true, y_pred).numpy() 2.354 >>> # Using 'none' reduction type. >>> cce = tf.keras.losses.CategoricalCrossentropy( ... reduction=tf.keras.losses.Reduction.NONE) >>> cce(y_true, y_pred).numpy() array([0.0513, 2.303], dtype=float32) Usage with the `compile()` API: ```python model.compile(optimizer='sgd', loss=tf.keras.losses.CategoricalCrossentropy()) ``` FrrVcategorical_crossentropycstjt|||||ddS)a Initializes `CategoricalCrossentropy` instance. Args: from_logits: Whether `y_pred` is expected to be a logits tensor. By default, we assume that `y_pred` encodes a probability distribution. label_smoothing: Float in [0, 1]. When > 0, label values are smoothed, meaning the confidence on label values are relaxed. For example, if `0.1`, use `0.1 / num_classes` for non-target labels and `0.9 + 0.1 / num_classes` for target labels. axis: The axis along which to compute crossentropy (the features axis). Defaults to -1. reduction: Type of `tf.keras.losses.Reduction` to apply to loss. Default value is `AUTO`. `AUTO` indicates that the reduction option will be determined by the usage context. For almost all cases this defaults to `SUM_OVER_BATCH_SIZE`. When used with `tf.distribute.Strategy`, outside of built-in training loops such as `tf.keras` `compile` and `fit`, using `AUTO` or `SUM_OVER_BATCH_SIZE` will raise an error. Please see this custom training [tutorial]( https://www.tensorflow.org/tutorials/distribute/custom_training) for more details. name: Optional name for the instance. Defaults to 'categorical_crossentropy'. )rrrXrYrZN)rBr%r\)r"rXrYrZrr)r(r#r$r%sz CategoricalCrossentropy.__init__) r)r;r<r=r rr9r%rLr#r#)r(r$r[bs +r[z*keras.losses.SparseCategoricalCrossentropycs,eZdZdZdejjdffdd ZZS)SparseCategoricalCrossentropyavComputes the crossentropy loss between the labels and predictions. Use this crossentropy loss function when there are two or more label classes. We expect labels to be provided as integers. If you want to provide labels using `one-hot` representation, please use `CategoricalCrossentropy` loss. There should be `# classes` floating point values per feature for `y_pred` and a single floating point value per feature for `y_true`. In the snippet below, there is a single floating point value per example for `y_true` and `# classes` floating pointing values per example for `y_pred`. The shape of `y_true` is `[batch_size]` and the shape of `y_pred` is `[batch_size, num_classes]`. Standalone usage: >>> y_true = [1, 2] >>> y_pred = [[0.05, 0.95, 0], [0.1, 0.8, 0.1]] >>> # Using 'auto'/'sum_over_batch_size' reduction type. >>> scce = tf.keras.losses.SparseCategoricalCrossentropy() >>> scce(y_true, y_pred).numpy() 1.177 >>> # Calling with 'sample_weight'. >>> scce(y_true, y_pred, sample_weight=tf.constant([0.3, 0.7])).numpy() 0.814 >>> # Using 'sum' reduction type. >>> scce = tf.keras.losses.SparseCategoricalCrossentropy( ... reduction=tf.keras.losses.Reduction.SUM) >>> scce(y_true, y_pred).numpy() 2.354 >>> # Using 'none' reduction type. >>> scce = tf.keras.losses.SparseCategoricalCrossentropy( ... reduction=tf.keras.losses.Reduction.NONE) >>> scce(y_true, y_pred).numpy() array([0.0513, 2.303], dtype=float32) Usage with the `compile()` API: ```python model.compile(optimizer='sgd', loss=tf.keras.losses.SparseCategoricalCrossentropy()) ``` Fsparse_categorical_crossentropycstjt|||ddS)aInitializes `SparseCategoricalCrossentropy` instance. Args: from_logits: Whether `y_pred` is expected to be a logits tensor. By default, we assume that `y_pred` encodes a probability distribution. reduction: Type of `tf.keras.losses.Reduction` to apply to loss. Default value is `AUTO`. `AUTO` indicates that the reduction option will be determined by the usage context. For almost all cases this defaults to `SUM_OVER_BATCH_SIZE`. When used with `tf.distribute.Strategy`, outside of built-in training loops such as `tf.keras` `compile` and `fit`, using `AUTO` or `SUM_OVER_BATCH_SIZE` will raise an error. Please see this custom training [tutorial]( https://www.tensorflow.org/tutorials/distribute/custom_training) for more details. name: Optional name for the instance. Defaults to 'sparse_categorical_crossentropy'. )rrrXN)rBr%r^)r"rXrr)r(r#r$r%s z&SparseCategoricalCrossentropy.__init__) r)r;r<r=r rr9r%rLr#r#)r(r$r]s.r]zkeras.losses.Hingecs*eZdZdZejjdffdd ZZS)HingeaComputes the hinge loss between `y_true` and `y_pred`. `loss = maximum(1 - y_true * y_pred, 0)` `y_true` values are expected to be -1 or 1. If binary (0 or 1) labels are provided we will convert them to -1 or 1. Standalone usage: >>> y_true = [[0., 1.], [0., 0.]] >>> y_pred = [[0.6, 0.4], [0.4, 0.6]] >>> # Using 'auto'/'sum_over_batch_size' reduction type. >>> h = tf.keras.losses.Hinge() >>> h(y_true, y_pred).numpy() 1.3 >>> # Calling with 'sample_weight'. >>> h(y_true, y_pred, sample_weight=[1, 0]).numpy() 0.55 >>> # Using 'sum' reduction type. >>> h = tf.keras.losses.Hinge( ... reduction=tf.keras.losses.Reduction.SUM) >>> h(y_true, y_pred).numpy() 2.6 >>> # Using 'none' reduction type. >>> h = tf.keras.losses.Hinge( ... reduction=tf.keras.losses.Reduction.NONE) >>> h(y_true, y_pred).numpy() array([1.1, 1.5], dtype=float32) Usage with the `compile()` API: ```python model.compile(optimizer='sgd', loss=tf.keras.losses.Hinge()) ``` hingecstjt||ddS)aInitializes `Hinge` instance. Args: reduction: Type of `tf.keras.losses.Reduction` to apply to loss. Default value is `AUTO`. `AUTO` indicates that the reduction option will be determined by the usage context. For almost all cases this defaults to `SUM_OVER_BATCH_SIZE`. When used with `tf.distribute.Strategy`, outside of built-in training loops such as `tf.keras` `compile` and `fit`, using `AUTO` or `SUM_OVER_BATCH_SIZE` will raise an error. Please see this custom training [tutorial]( https://www.tensorflow.org/tutorials/distribute/custom_training) for more details. name: Optional name for the instance. Defaults to 'hinge'. )rrN)rBr%r`)r"rr)r(r#r$r%*szHinge.__init__) r)r;r<r=r rr9r%rLr#r#)r(r$r_s'r_zkeras.losses.SquaredHingecs*eZdZdZejjdffdd ZZS) SquaredHingea)Computes the squared hinge loss between `y_true` and `y_pred`. `loss = square(maximum(1 - y_true * y_pred, 0))` `y_true` values are expected to be -1 or 1. If binary (0 or 1) labels are provided we will convert them to -1 or 1. Standalone usage: >>> y_true = [[0., 1.], [0., 0.]] >>> y_pred = [[0.6, 0.4], [0.4, 0.6]] >>> # Using 'auto'/'sum_over_batch_size' reduction type. >>> h = tf.keras.losses.SquaredHinge() >>> h(y_true, y_pred).numpy() 1.86 >>> # Calling with 'sample_weight'. >>> h(y_true, y_pred, sample_weight=[1, 0]).numpy() 0.73 >>> # Using 'sum' reduction type. >>> h = tf.keras.losses.SquaredHinge( ... reduction=tf.keras.losses.Reduction.SUM) >>> h(y_true, y_pred).numpy() 3.72 >>> # Using 'none' reduction type. >>> h = tf.keras.losses.SquaredHinge( ... reduction=tf.keras.losses.Reduction.NONE) >>> h(y_true, y_pred).numpy() array([1.46, 2.26], dtype=float32) Usage with the `compile()` API: ```python model.compile(optimizer='sgd', loss=tf.keras.losses.SquaredHinge()) ``` squared_hingecstjt||ddS)aInitializes `SquaredHinge` instance. Args: reduction: Type of `tf.keras.losses.Reduction` to apply to loss. Default value is `AUTO`. `AUTO` indicates that the reduction option will be determined by the usage context. For almost all cases this defaults to `SUM_OVER_BATCH_SIZE`. When used with `tf.distribute.Strategy`, outside of built-in training loops such as `tf.keras` `compile` and `fit`, using `AUTO` or `SUM_OVER_BATCH_SIZE` will raise an error. Please see this custom training [tutorial]( https://www.tensorflow.org/tutorials/distribute/custom_training) for more details. name: Optional name for the instance. Defaults to 'squared_hinge'. )rrN)rBr%rb)r"rr)r(r#r$r%eszSquaredHinge.__init__) r)r;r<r=r rr9r%rLr#r#)r(r$ra<s'razkeras.losses.CategoricalHingecs*eZdZdZejjdffdd ZZS)CategoricalHingeaComputes the categorical hinge loss between `y_true` and `y_pred`. `loss = maximum(neg - pos + 1, 0)` where `neg=maximum((1-y_true)*y_pred) and pos=sum(y_true*y_pred)` Standalone usage: >>> y_true = [[0, 1], [0, 0]] >>> y_pred = [[0.6, 0.4], [0.4, 0.6]] >>> # Using 'auto'/'sum_over_batch_size' reduction type. >>> h = tf.keras.losses.CategoricalHinge() >>> h(y_true, y_pred).numpy() 1.4 >>> # Calling with 'sample_weight'. >>> h(y_true, y_pred, sample_weight=[1, 0]).numpy() 0.6 >>> # Using 'sum' reduction type. >>> h = tf.keras.losses.CategoricalHinge( ... reduction=tf.keras.losses.Reduction.SUM) >>> h(y_true, y_pred).numpy() 2.8 >>> # Using 'none' reduction type. >>> h = tf.keras.losses.CategoricalHinge( ... reduction=tf.keras.losses.Reduction.NONE) >>> h(y_true, y_pred).numpy() array([1.2, 1.6], dtype=float32) Usage with the `compile()` API: ```python model.compile(optimizer='sgd', loss=tf.keras.losses.CategoricalHinge()) ``` categorical_hingecstjt||ddS)aInitializes `CategoricalHinge` instance. Args: reduction: Type of `tf.keras.losses.Reduction` to apply to loss. Default value is `AUTO`. `AUTO` indicates that the reduction option will be determined by the usage context. For almost all cases this defaults to `SUM_OVER_BATCH_SIZE`. When used with `tf.distribute.Strategy`, outside of built-in training loops such as `tf.keras` `compile` and `fit`, using `AUTO` or `SUM_OVER_BATCH_SIZE` will raise an error. Please see this custom training [tutorial]( https://www.tensorflow.org/tutorials/distribute/custom_training) for more details. name: Optional name for the instance. Defaults to 'categorical_hinge'. )rrN)rBr%rd)r"rr)r(r#r$r%szCategoricalHinge.__init__) r)r;r<r=r rr9r%rLr#r#)r(r$rcys%rczkeras.losses.Poissoncs*eZdZdZejjdffdd ZZS)PoissonaComputes the Poisson loss between `y_true` and `y_pred`. `loss = y_pred - y_true * log(y_pred)` Standalone usage: >>> y_true = [[0., 1.], [0., 0.]] >>> y_pred = [[1., 1.], [0., 0.]] >>> # Using 'auto'/'sum_over_batch_size' reduction type. >>> p = tf.keras.losses.Poisson() >>> p(y_true, y_pred).numpy() 0.5 >>> # Calling with 'sample_weight'. >>> p(y_true, y_pred, sample_weight=[0.8, 0.2]).numpy() 0.4 >>> # Using 'sum' reduction type. >>> p = tf.keras.losses.Poisson( ... reduction=tf.keras.losses.Reduction.SUM) >>> p(y_true, y_pred).numpy() 0.999 >>> # Using 'none' reduction type. >>> p = tf.keras.losses.Poisson( ... reduction=tf.keras.losses.Reduction.NONE) >>> p(y_true, y_pred).numpy() array([0.999, 0.], dtype=float32) Usage with the `compile()` API: ```python model.compile(optimizer='sgd', loss=tf.keras.losses.Poisson()) ``` poissoncstjt||ddS)aInitializes `Poisson` instance. Args: reduction: Type of `tf.keras.losses.Reduction` to apply to loss. Default value is `AUTO`. `AUTO` indicates that the reduction option will be determined by the usage context. For almost all cases this defaults to `SUM_OVER_BATCH_SIZE`. When used with `tf.distribute.Strategy`, outside of built-in training loops such as `tf.keras` `compile` and `fit`, using `AUTO` or `SUM_OVER_BATCH_SIZE` will raise an error. Please see this custom training [tutorial]( https://www.tensorflow.org/tutorials/distribute/custom_training) for more details. name: Optional name for the instance. Defaults to 'poisson'. )rrN)rBr%rf)r"rr)r(r#r$r%szPoisson.__init__) r)r;r<r=r rr9r%rLr#r#)r(r$res$rezkeras.losses.LogCoshcs*eZdZdZejjdffdd ZZS)LogCoshaComputes the logarithm of the hyperbolic cosine of the prediction error. `logcosh = log((exp(x) + exp(-x))/2)`, where x is the error `y_pred - y_true`. Standalone usage: >>> y_true = [[0., 1.], [0., 0.]] >>> y_pred = [[1., 1.], [0., 0.]] >>> # Using 'auto'/'sum_over_batch_size' reduction type. >>> l = tf.keras.losses.LogCosh() >>> l(y_true, y_pred).numpy() 0.108 >>> # Calling with 'sample_weight'. >>> l(y_true, y_pred, sample_weight=[0.8, 0.2]).numpy() 0.087 >>> # Using 'sum' reduction type. >>> l = tf.keras.losses.LogCosh( ... reduction=tf.keras.losses.Reduction.SUM) >>> l(y_true, y_pred).numpy() 0.217 >>> # Using 'none' reduction type. >>> l = tf.keras.losses.LogCosh( ... reduction=tf.keras.losses.Reduction.NONE) >>> l(y_true, y_pred).numpy() array([0.217, 0.], dtype=float32) Usage with the `compile()` API: ```python model.compile(optimizer='sgd', loss=tf.keras.losses.LogCosh()) ``` log_coshcstjt||ddS)aInitializes `LogCosh` instance. Args: reduction: Type of `tf.keras.losses.Reduction` to apply to loss. Default value is `AUTO`. `AUTO` indicates that the reduction option will be determined by the usage context. For almost all cases this defaults to `SUM_OVER_BATCH_SIZE`. When used with `tf.distribute.Strategy`, outside of built-in training loops such as `tf.keras` `compile` and `fit`, using `AUTO` or `SUM_OVER_BATCH_SIZE` will raise an error. Please see this custom training [tutorial]( https://www.tensorflow.org/tutorials/distribute/custom_training) for more details. name: Optional name for the instance. Defaults to 'log_cosh'. )rrN)rBr%rh)r"rr)r(r#r$r%szLogCosh.__init__) r)r;r<r=r rr9r%rLr#r#)r(r$rgs%rgzkeras.losses.KLDivergencecs*eZdZdZejjdffdd ZZS) KLDivergencea Computes Kullback-Leibler divergence loss between `y_true` and `y_pred`. `loss = y_true * log(y_true / y_pred)` See: https://en.wikipedia.org/wiki/Kullback%E2%80%93Leibler_divergence Standalone usage: >>> y_true = [[0, 1], [0, 0]] >>> y_pred = [[0.6, 0.4], [0.4, 0.6]] >>> # Using 'auto'/'sum_over_batch_size' reduction type. >>> kl = tf.keras.losses.KLDivergence() >>> kl(y_true, y_pred).numpy() 0.458 >>> # Calling with 'sample_weight'. >>> kl(y_true, y_pred, sample_weight=[0.8, 0.2]).numpy() 0.366 >>> # Using 'sum' reduction type. >>> kl = tf.keras.losses.KLDivergence( ... reduction=tf.keras.losses.Reduction.SUM) >>> kl(y_true, y_pred).numpy() 0.916 >>> # Using 'none' reduction type. >>> kl = tf.keras.losses.KLDivergence( ... reduction=tf.keras.losses.Reduction.NONE) >>> kl(y_true, y_pred).numpy() array([0.916, -3.08e-06], dtype=float32) Usage with the `compile()` API: ```python model.compile(optimizer='sgd', loss=tf.keras.losses.KLDivergence()) ``` kl_divergencecstjt||ddS)aInitializes `KLDivergence` instance. Args: reduction: Type of `tf.keras.losses.Reduction` to apply to loss. Default value is `AUTO`. `AUTO` indicates that the reduction option will be determined by the usage context. For almost all cases this defaults to `SUM_OVER_BATCH_SIZE`. When used with `tf.distribute.Strategy`, outside of built-in training loops such as `tf.keras` `compile` and `fit`, using `AUTO` or `SUM_OVER_BATCH_SIZE` will raise an error. Please see this custom training [tutorial]( https://www.tensorflow.org/tutorials/distribute/custom_training) for more details. name: Optional name for the instance. Defaults to 'kl_divergence'. )rrN)rBr%rj)r"rr)r(r#r$r%MszKLDivergence.__init__) r)r;r<r=r rr9r%rLr#r#)r(r$ri%s&rizkeras.losses.Hubercs,eZdZdZdejjdffdd ZZS)Hubera7Computes the Huber loss between `y_true` and `y_pred`. For each value x in `error = y_true - y_pred`: ``` loss = 0.5 * x^2 if |x| <= d loss = 0.5 * d^2 + d * (|x| - d) if |x| > d ``` where d is `delta`. See: https://en.wikipedia.org/wiki/Huber_loss Standalone usage: >>> y_true = [[0, 1], [0, 0]] >>> y_pred = [[0.6, 0.4], [0.4, 0.6]] >>> # Using 'auto'/'sum_over_batch_size' reduction type. >>> h = tf.keras.losses.Huber() >>> h(y_true, y_pred).numpy() 0.155 >>> # Calling with 'sample_weight'. >>> h(y_true, y_pred, sample_weight=[1, 0]).numpy() 0.09 >>> # Using 'sum' reduction type. >>> h = tf.keras.losses.Huber( ... reduction=tf.keras.losses.Reduction.SUM) >>> h(y_true, y_pred).numpy() 0.31 >>> # Using 'none' reduction type. >>> h = tf.keras.losses.Huber( ... reduction=tf.keras.losses.Reduction.NONE) >>> h(y_true, y_pred).numpy() array([0.18, 0.13], dtype=float32) Usage with the `compile()` API: ```python model.compile(optimizer='sgd', loss=tf.keras.losses.Huber()) ``` g? huber_losscstjt|||ddS)aBInitializes `Huber` instance. Args: delta: A float, the point where the Huber loss function changes from a quadratic to linear. reduction: Type of `tf.keras.losses.Reduction` to apply to loss. Default value is `AUTO`. `AUTO` indicates that the reduction option will be determined by the usage context. For almost all cases this defaults to `SUM_OVER_BATCH_SIZE`. When used with `tf.distribute.Strategy`, outside of built-in training loops such as `tf.keras` `compile` and `fit`, using `AUTO` or `SUM_OVER_BATCH_SIZE` will raise an error. Please see this custom training [tutorial]( https://www.tensorflow.org/tutorials/distribute/custom_training) for more details. name: Optional name for the instance. Defaults to 'huber_loss'. )rrdeltaN)rBr%huber)r"rmrr)r(r#r$r%szHuber.__init__) r)r;r<r=r rr9r%rLr#r#)r(r$rkas*rkz keras.metrics.mean_squared_errorzkeras.metrics.msezkeras.metrics.MSEzkeras.losses.mean_squared_errorzkeras.losses.msezkeras.losses.MSEcCs.t|}t||j}tjt||ddS)aComputes the mean squared error between labels and predictions. After computing the squared distance between the inputs, the mean value over the last dimension is returned. `loss = mean(square(y_true - y_pred), axis=-1)` Standalone usage: >>> y_true = np.random.randint(0, 2, size=(2, 3)) >>> y_pred = np.random.random(size=(2, 3)) >>> loss = tf.keras.losses.mean_squared_error(y_true, y_pred) >>> assert loss.shape == (2,) >>> assert np.array_equal( ... loss.numpy(), np.mean(np.square(y_true - y_pred), axis=-1)) Args: y_true: Ground truth values. shape = `[batch_size, d0, .. dN]`. y_pred: The predicted values. shape = `[batch_size, d0, .. dN]`. Returns: Mean squared error values. shape = `[batch_size, d0, .. dN-1]`. rV)rZ)r"convert_to_tensor_v2_with_dispatchrcastdtyper meansquared_difference)r1r2r#r#r$rNs rNFc sddddfddfdd}t|tjsH||S|jd d }t|d krxtj||jd }nt j g|jd }d d||fD}|rdd|D}|d |d d kr|d dd |d <t j |t|d kd} t |} t| tj| ||f|dSQRXdS)aApply a loss function on a per batch basis. Args: loss_fn: The loss function y_true: truth values (RaggedTensor) y_pred: predicted values (RaggedTensor) y_pred_extra_dim: whether y_pred has an additional dimension compared to y_true Returns: Loss-function result. A dense tensor if the output has a single dimension (per-batch loss value); a ragged tensor otherwise. cSstdd|DS)zReturns true if this RaggedTensor has the same row_lenghts across all ragged dimensions and thus can be converted to a dense tensor without loss of information. Args: rt: RaggedTensor. c Ss2g|]*}ttt|ttdgqS)g)requalZreduce_variancerpr floatxrZconstant).0Zrow_lensr#r#r$ szH_ragged_tensor_apply_loss..rt_is_equiv_dense..)r reduce_allZnested_row_lengths)rtr#r#r$rt_is_equiv_denses z4_ragged_tensor_apply_loss..rt_is_equiv_densecSstdd|DS)Ncss&|]}t|tjr|n|VqdS)N) isinstancer RaggedTensor to_tensor)rvryr#r#r$ szG_ragged_tensor_apply_loss.._convert_to_dense..)tuple)inputsr#r#r$_convert_to_densesz4_ragged_tensor_apply_loss.._convert_to_densecsB|}|r&t|tjs&tj|}n|s>t|tjr>|}|S)z Adapt the result to ragged or dense tensor according to the expected output type. This is done so that all the return values of the map operation have the same type. )r{rr|Z from_tensorr})r ragged_outputr)loss_fnr#r$ _call_losss z-_ragged_tensor_apply_loss.._call_losscsH\}}t|tjr@t|fddfddSS)NcsS)Nr#r#)rrrrr#r$z=_ragged_tensor_apply_loss.._wrapper..cs S)Nr#r#)rrrr#r$rr)r{rr|rZcond)rrr'r2)rrrrz)rrr$_wrappers z+_ragged_tensor_apply_loss.._wrapperrVr)shaperqcSsg|] }|jqSr#)Znested_row_splits)rvryr#r#r$rwsz-_ragged_tensor_apply_loss..cSsg|] }t|qSr#)len)rvslistr#r#r$rw sN)r)Zelemsrq)r{rr|r}ras_listrZRaggedTensorSpecrqr Z TensorSpec functoolspartialrZassert_splits_matchrZcontrol_dependenciesrmap_fn) rr1r2y_pred_extra_dimrZlshapespecZnested_splits_listZrdimsrZassertion_listr#)rrrrzr$_ragged_tensor_apply_losss&     rcCs tt||S)aImplements support for handling RaggedTensors. Args: y_true: RaggedTensor truth values. shape = `[batch_size, d0, .. dN]`. y_pred: RaggedTensor predicted values. shape = `[batch_size, d0, .. dN]`. Returns: Mean squared error values. shape = `[batch_size, d0, .. dN-1]`. When the number of dimensions of the batch feature vector [d0, .. dN] is greater than one the return value is a RaggedTensor. Otherwise a Dense tensor with dimensions [batch_size] is returned. )rrN)r1r2r#r#r$_ragged_tensor_msesrz!keras.metrics.mean_absolute_errorzkeras.metrics.maezkeras.metrics.MAEz keras.losses.mean_absolute_errorzkeras.losses.maezkeras.losses.MAEcCs0t|}t||j}tjt||ddS)aComputes the mean absolute error between labels and predictions. `loss = mean(abs(y_true - y_pred), axis=-1)` Standalone usage: >>> y_true = np.random.randint(0, 2, size=(2, 3)) >>> y_pred = np.random.random(size=(2, 3)) >>> loss = tf.keras.losses.mean_absolute_error(y_true, y_pred) >>> assert loss.shape == (2,) >>> assert np.array_equal( ... loss.numpy(), np.mean(np.abs(y_true - y_pred), axis=-1)) Args: y_true: Ground truth values. shape = `[batch_size, d0, .. dN]`. y_pred: The predicted values. shape = `[batch_size, d0, .. dN]`. Returns: Mean absolute error values. shape = `[batch_size, d0, .. dN-1]`. rV)rZ)rrorrprqr rrabs)r1r2r#r#r$rP's rPcCs tt||S)z-RaggedTensor adapter for mean_absolute_error.)rrP)r1r2r#r#r$_ragged_tensor_maeEsrz,keras.metrics.mean_absolute_percentage_errorzkeras.metrics.mapezkeras.metrics.MAPEz+keras.losses.mean_absolute_percentage_errorzkeras.losses.mapezkeras.losses.MAPEcCsNt|}t||j}t||tt|t}dtj |ddS)aComputes the mean absolute percentage error between `y_true` and `y_pred`. `loss = 100 * mean(abs((y_true - y_pred) / y_true), axis=-1)` Standalone usage: >>> y_true = np.random.random(size=(2, 3)) >>> y_true = np.maximum(y_true, 1e-7) # Prevent division by zero >>> y_pred = np.random.random(size=(2, 3)) >>> loss = tf.keras.losses.mean_absolute_percentage_error(y_true, y_pred) >>> assert loss.shape == (2,) >>> assert np.array_equal( ... loss.numpy(), ... 100. * np.mean(np.abs((y_true - y_pred) / y_true), axis=-1)) Args: y_true: Ground truth values. shape = `[batch_size, d0, .. dN]`. y_pred: The predicted values. shape = `[batch_size, d0, .. dN]`. Returns: Mean absolute percentage error values. shape = `[batch_size, d0, .. dN-1]`. gY@rV)rZ) rrorrprqrr maximumepsilonrr)r1r2diffr#r#r$rRKs  rRcCs tt||S)zSupport RaggedTensors.)rrR)r1r2r#r#r$_ragged_tensor_mapeosrz,keras.metrics.mean_squared_logarithmic_errorzkeras.metrics.mslezkeras.metrics.MSLEz+keras.losses.mean_squared_logarithmic_errorzkeras.losses.mslezkeras.losses.MSLEcCsbt|}t||j}tt|td}tt|td}tj t ||ddS)aFComputes the mean squared logarithmic error between `y_true` and `y_pred`. `loss = mean(square(log(y_true + 1) - log(y_pred + 1)), axis=-1)` Standalone usage: >>> y_true = np.random.randint(0, 2, size=(2, 3)) >>> y_pred = np.random.random(size=(2, 3)) >>> loss = tf.keras.losses.mean_squared_logarithmic_error(y_true, y_pred) >>> assert loss.shape == (2,) >>> y_true = np.maximum(y_true, 1e-7) >>> y_pred = np.maximum(y_pred, 1e-7) >>> assert np.allclose( ... loss.numpy(), ... np.mean( ... np.square(np.log(y_true + 1.) - np.log(y_pred + 1.)), axis=-1)) Args: y_true: Ground truth values. shape = `[batch_size, d0, .. dN]`. y_pred: The predicted values. shape = `[batch_size, d0, .. dN]`. Returns: Mean squared logarithmic error values. shape = `[batch_size, d0, .. dN-1]`. g?rV)rZ) rrorrprqlogr rrrrrs)r1r2Z first_logZ second_logr#r#r$rTws  rTcCs tt||S)z.Implements support for handling RaggedTensors.)rrT)r1r2r#r#r$_ragged_tensor_mslesrcsPtd}td}tt||}fdd}t||fdd}|S)z!Converts binary labels into -1/1.rrcs ddS)Ng@g?r#r#)r1r#r$_convert_binary_labelssz5_maybe_convert_labels.._convert_binary_labelscsS)Nr#r#)r1r#r$rrz'_maybe_convert_labels..)rrtrx logical_orr)r1Z are_zerosZare_onesZ is_binaryrZupdated_y_truer#)r1r$_maybe_convert_labelss   rzkeras.metrics.squared_hingezkeras.losses.squared_hingecCsDt|}t||j}t|}tjtt d||dddS)aAComputes the squared hinge loss between `y_true` and `y_pred`. `loss = mean(square(maximum(1 - y_true * y_pred, 0)), axis=-1)` Standalone usage: >>> y_true = np.random.choice([-1, 1], size=(2, 3)) >>> y_pred = np.random.random(size=(2, 3)) >>> loss = tf.keras.losses.squared_hinge(y_true, y_pred) >>> assert loss.shape == (2,) >>> assert np.array_equal( ... loss.numpy(), ... np.mean(np.square(np.maximum(1. - y_true * y_pred, 0.)), axis=-1)) Args: y_true: The ground truth values. `y_true` values are expected to be -1 or 1. If binary (0 or 1) labels are provided we will convert them to -1 or 1. shape = `[batch_size, d0, .. dN]`. y_pred: The predicted values. shape = `[batch_size, d0, .. dN]`. Returns: Squared hinge loss values. shape = `[batch_size, d0, .. dN-1]`. g?grV)rZ) rrorrprqrr rrsquarer)r1r2r#r#r$rbs  rbzkeras.metrics.hingezkeras.losses.hingecCs>t|}t||j}t|}tjtd||dddS)aComputes the hinge loss between `y_true` and `y_pred`. `loss = mean(maximum(1 - y_true * y_pred, 0), axis=-1)` Standalone usage: >>> y_true = np.random.choice([-1, 1], size=(2, 3)) >>> y_pred = np.random.random(size=(2, 3)) >>> loss = tf.keras.losses.hinge(y_true, y_pred) >>> assert loss.shape == (2,) >>> assert np.array_equal( ... loss.numpy(), ... np.mean(np.maximum(1. - y_true * y_pred, 0.), axis=-1)) Args: y_true: The ground truth values. `y_true` values are expected to be -1 or 1. If binary (0 or 1) labels are provided they will be converted to -1 or 1. shape = `[batch_size, d0, .. dN]`. y_pred: The predicted values. shape = `[batch_size, d0, .. dN]`. Returns: Hinge loss values. shape = `[batch_size, d0, .. dN-1]`. g?grV)rZ) rrorrprqrr rrr)r1r2r#r#r$r`s r`zkeras.losses.categorical_hingecCsbt|}t||j}tj||dd}tjd||dd}td|j}t||d|S)aXComputes the categorical hinge loss between `y_true` and `y_pred`. `loss = maximum(neg - pos + 1, 0)` where `neg=maximum((1-y_true)*y_pred) and pos=sum(y_true*y_pred)` Standalone usage: >>> y_true = np.random.randint(0, 3, size=(2,)) >>> y_true = tf.keras.utils.to_categorical(y_true, num_classes=3) >>> y_pred = np.random.random(size=(2, 3)) >>> loss = tf.keras.losses.categorical_hinge(y_true, y_pred) >>> assert loss.shape == (2,) >>> pos = np.sum(y_true * y_pred, axis=-1) >>> neg = np.amax((1. - y_true) * y_pred, axis=-1) >>> assert np.array_equal(loss.numpy(), np.maximum(0., neg - pos + 1.)) Args: y_true: The ground truth values. `y_true` values are expected to be either `{-1, +1}` or `{0, 1}` (i.e. a one-hot-encoded tensor). y_pred: The predicted values. Returns: Categorical hinge loss values. rV)rZg?g)rrorrprq reduce_sumZ reduce_maxr)r1r2posnegzeror#r#r$rds  rdzkeras.losses.huber)Zv1?c Cstj|td}tj|td}tj|td}t||}t|}tjd|jd}tj t ||k|t ||||t |ddS)aComputes Huber loss value. For each value x in `error = y_true - y_pred`: ``` loss = 0.5 * x^2 if |x| <= d loss = d * |x| - 0.5 * d^2 if |x| > d ``` where d is `delta`. See: https://en.wikipedia.org/wiki/Huber_loss Args: y_true: tensor of true targets. y_pred: tensor of predicted targets. delta: A float, the point where the Huber loss function changes from a quadratic to linear. Returns: Tensor with one scalar loss entry per sample. )rqg?rV)rZ) rrpr rusubtractrrrorqrrrZwhere_v2r)r1r2rmerrorZ abs_errorZhalfr#r#r$rns  rnzkeras.losses.log_coshzkeras.losses.logcoshzkeras.metrics.log_coshzkeras.metrics.logcoshcCs6t|}t||j}dd}tj|||ddS)aLogarithm of the hyperbolic cosine of the prediction error. `log(cosh(x))` is approximately equal to `(x ** 2) / 2` for small `x` and to `abs(x) - log(2)` for large `x`. This means that 'logcosh' works mostly like the mean squared error, but will not be so strongly affected by the occasional wildly incorrect prediction. Standalone usage: >>> y_true = np.random.random(size=(2, 3)) >>> y_pred = np.random.random(size=(2, 3)) >>> loss = tf.keras.losses.logcosh(y_true, y_pred) >>> assert loss.shape == (2,) >>> x = y_pred - y_true >>> assert np.allclose( ... loss.numpy(), ... np.mean(x + np.log(np.exp(-2. * x) + 1.) - math_ops.log(2.), axis=-1), ... atol=1e-5) Args: y_true: Ground truth values. shape = `[batch_size, d0, .. dN]`. y_pred: The predicted values. shape = `[batch_size, d0, .. dN]`. Returns: Logcosh error values. shape = `[batch_size, d0, .. dN-1]`. cSs&|td|ttd|jS)Ngg@)rZsoftplusrprrq)xr#r#r$_logcosh[szlog_cosh.._logcoshrV)rZ)rrorrprqr rr)r1r2rr#r#r$rh:s rhz&keras.metrics.categorical_crossentropyz%keras.losses.categorical_crossentropyrVcsbttjtjtdfdd}t|fddtj||dS)aComputes the categorical crossentropy loss. Standalone usage: >>> y_true = [[0, 1, 0], [0, 0, 1]] >>> y_pred = [[0.05, 0.95, 0], [0.1, 0.8, 0.1]] >>> loss = tf.keras.losses.categorical_crossentropy(y_true, y_pred) >>> assert loss.shape == (2,) >>> loss.numpy() array([0.0513, 2.303], dtype=float32) Args: y_true: Tensor of one-hot true targets. y_pred: Tensor of predicted targets. from_logits: Whether `y_pred` is expected to be a logits tensor. By default, we assume that `y_pred` encodes a probability distribution. label_smoothing: Float in [0, 1]. If > `0` then smooth the labels. For example, if `0.1`, use `0.1 / num_classes` for non-target labels and `0.9 + 0.1 / num_classes` for target labels. axis: Defaults to -1. The dimension along which the entropy is computed. Returns: Categorical crossentropy loss value. )rqcs,ttdj}d|S)NrVg?)rrprrrq)Z num_classes)rYr2r1r#r$_smooth_labelssz0categorical_crossentropy.._smooth_labelscsS)Nr#r#)r1r#r$rrz*categorical_crossentropy..)rXrZ) rrorrprqr rurr\)r1r2rXrYrZrr#)rYr2r1r$r\bs! r\cCstjt|||d}t|||S)aWImplements support for handling RaggedTensors. Args: y_true: Tensor of one-hot true targets. y_pred: Tensor of predicted targets. from_logits: Whether `y_pred` is expected to be a logits tensor. By default, we assume that `y_pred` encodes a probability distribution. label_smoothing: Float in [0, 1]. If > `0` then smooth the labels. For example, if `0.1`, use `0.1 / num_classes` for non-target labels and `0.9 + 0.1 / num_classes` for target labels. axis: The axis along which to compute crossentropy (the features axis). Defaults to -1. Returns: Categorical crossentropy loss value. Expected shape: (batch, sequence_len, n_classes) with sequence_len being variable per batch. Return shape: (batch, sequence_len). When used by CategoricalCrossentropy() with the default reduction (SUM_OVER_BATCH_SIZE), the reduction averages the loss over the number of elements independent of the batch. E.g. if the RaggedTensor has 2 batches with [2, 1] values respectivly the resulting loss is the sum of the individual loss values divided by 3. )rXrYrZ)rrr\r)r1r2rXrYrZrCr#r#r$'_ragged_tensor_categorical_crossentropys !rz-keras.metrics.sparse_categorical_crossentropyz,keras.losses.sparse_categorical_crossentropycCs*t|}t||j}tj||||dS)aComputes the sparse categorical crossentropy loss. Standalone usage: >>> y_true = [1, 2] >>> y_pred = [[0.05, 0.95, 0], [0.1, 0.8, 0.1]] >>> loss = tf.keras.losses.sparse_categorical_crossentropy(y_true, y_pred) >>> assert loss.shape == (2,) >>> loss.numpy() array([0.0513, 2.303], dtype=float32) Args: y_true: Ground truth values. y_pred: The predicted values. from_logits: Whether `y_pred` is expected to be a logits tensor. By default, we assume that `y_pred` encodes a probability distribution. axis: Defaults to -1. The dimension along which the entropy is computed. Returns: Sparse categorical crossentropy loss value. )rXrZ)rrorrprqr r^)r1r2rXrZr#r#r$r^s r^cCs tjt||d}t|||ddS)a4 Implements support for handling RaggedTensors. Expected y_pred shape: (batch, sequence_len, n_classes) with sequence_len being variable per batch. Return shape: (batch, sequence_len). When used by SparseCategoricalCrossentropy() with the default reduction (SUM_OVER_BATCH_SIZE), the reduction averages the loss over the number of elements independent of the batch. E.g. if the RaggedTensor has 2 batches with [2, 1] values respectively, the resulting loss is the sum of the individual loss values divided by 3. )rXrZT)r)rrr^r)r1r2rXrZrCr#r#r$._ragged_tensor_sparse_categorical_crossentropys rz!keras.metrics.binary_crossentropyz keras.losses.binary_crossentropycsht|}t|jtjtdfdd}t|fddtjtj ||d|dS)aComputes the binary crossentropy loss. Standalone usage: >>> y_true = [[0, 1], [0, 0]] >>> y_pred = [[0.6, 0.4], [0.4, 0.6]] >>> loss = tf.keras.losses.binary_crossentropy(y_true, y_pred) >>> assert loss.shape == (2,) >>> loss.numpy() array([0.916 , 0.714], dtype=float32) Args: y_true: Ground truth values. shape = `[batch_size, d0, .. dN]`. y_pred: The predicted values. shape = `[batch_size, d0, .. dN]`. from_logits: Whether `y_pred` is expected to be a logits tensor. By default, we assume that `y_pred` encodes a probability distribution. label_smoothing: Float in [0, 1]. If > `0` then smooth the labels by squeezing them towards 0.5 That is, using `1. - 0.5 * label_smoothing` for the target class and `0.5 * label_smoothing` for the non-target class. axis: The axis along which the mean is computed. Defaults to -1. Returns: Binary crossentropy loss value. shape = `[batch_size, d0, .. dN-1]`. )rqcsddS)Ng?g?r#r#)rYr1r#r$rsz+binary_crossentropy.._smooth_labelscsS)Nr#r#)r1r#r$rrz%binary_crossentropy..)rX)rZ) rrorrprqr rurrrrW)r1r2rXrYrZrr#)rYr1r$rWs rWcCstjt|||d}t|||S)aImplements support for handling RaggedTensors. Args: y_true: Tensor of one-hot true targets. y_pred: Tensor of predicted targets. from_logits: Whether `y_pred` is expected to be a logits tensor. By default, we assume that `y_pred` encodes a probability distribution. label_smoothing: Float in [0, 1]. If > `0` then smooth the labels. For example, if `0.1`, use `0.1 / num_classes` for non-target labels and `0.9 + 0.1 / num_classes` for target labels. axis: Axis along which to compute crossentropy. Returns: Binary crossentropy loss value. Expected shape: (batch, sequence_len) with sequence_len being variable per batch. Return shape: (batch,); returns the per batch mean of the loss values. When used by BinaryCrossentropy() with the default reduction (SUM_OVER_BATCH_SIZE), the reduction averages the per batch losses over the number of batches. )rXrYrZ)rrrWr)r1r2rXrYrZrCr#r#r$"_ragged_tensor_binary_crossentropy#s rzkeras.metrics.kl_divergencez)keras.metrics.kullback_leibler_divergencezkeras.metrics.kldzkeras.metrics.KLDzkeras.losses.kl_divergencez(keras.losses.kullback_leibler_divergencezkeras.losses.kldzkeras.losses.KLDcCsXt|}t||j}t|td}t|td}tj|t ||ddS)aNComputes Kullback-Leibler divergence loss between `y_true` and `y_pred`. `loss = y_true * log(y_true / y_pred)` See: https://en.wikipedia.org/wiki/Kullback%E2%80%93Leibler_divergence Standalone usage: >>> y_true = np.random.randint(0, 2, size=(2, 3)).astype(np.float64) >>> y_pred = np.random.random(size=(2, 3)) >>> loss = tf.keras.losses.kullback_leibler_divergence(y_true, y_pred) >>> assert loss.shape == (2,) >>> y_true = tf.keras.backend.clip(y_true, 1e-7, 1) >>> y_pred = tf.keras.backend.clip(y_pred, 1e-7, 1) >>> assert np.array_equal( ... loss.numpy(), np.sum(y_true * np.log(y_true / y_pred), axis=-1)) Args: y_true: Tensor of true targets. y_pred: Tensor of predicted targets. Returns: A `Tensor` with loss. Raises: TypeError: If `y_true` cannot be cast to the `y_pred.dtype`. rrV)rZ) rrorrprqr Zcliprrr)r1r2r#r#r$rjHs " rjzkeras.metrics.poissonzkeras.losses.poissoncCs<t|}t||j}tj||t|tddS)a2Computes the Poisson loss between y_true and y_pred. The Poisson loss is the mean of the elements of the `Tensor` `y_pred - y_true * log(y_pred)`. Standalone usage: >>> y_true = np.random.randint(0, 2, size=(2, 3)) >>> y_pred = np.random.random(size=(2, 3)) >>> loss = tf.keras.losses.poisson(y_true, y_pred) >>> assert loss.shape == (2,) >>> y_pred = y_pred + 1e-7 >>> assert np.allclose( ... loss.numpy(), np.mean(y_pred - y_true * np.log(y_pred), axis=-1), ... atol=1e-5) Args: y_true: Ground truth values. shape = `[batch_size, d0, .. dN]`. y_pred: The predicted values. shape = `[batch_size, d0, .. dN]`. Returns: Poisson loss value. shape = `[batch_size, d0, .. dN-1]`. Raises: InvalidArgumentError: If `y_true` and `y_pred` have incompatible shapes. rV)rZ) rrorrprqr rrrr)r1r2r#r#r$rfqs rfzkeras.losses.cosine_similarityzkeras.metrics.cosine_proximityzkeras.metrics.cosinezkeras.losses.cosine_proximityzkeras.losses.cosinecCs0tj||d}tj||d}tj|||d S)aComputes the cosine similarity between labels and predictions. Note that it is a number between -1 and 1. When it is a negative number between -1 and 0, 0 indicates orthogonality and values closer to -1 indicate greater similarity. The values closer to 1 indicate greater dissimilarity. This makes it usable as a loss function in a setting where you try to maximize the proximity between predictions and targets. If either `y_true` or `y_pred` is a zero vector, cosine similarity will be 0 regardless of the proximity between predictions and targets. `loss = -sum(l2_norm(y_true) * l2_norm(y_pred))` Standalone usage: >>> y_true = [[0., 1.], [1., 1.], [1., 1.]] >>> y_pred = [[1., 0.], [1., 1.], [-1., -1.]] >>> loss = tf.keras.losses.cosine_similarity(y_true, y_pred, axis=1) >>> loss.numpy() array([-0., -0.999, 0.999], dtype=float32) Args: y_true: Tensor of true targets. y_pred: Tensor of predicted targets. axis: Axis along which to determine similarity. Returns: Cosine similarity tensor. )rZ)rZ l2_normalizerr)r1r2rZr#r#r$cosine_similaritys(rzkeras.losses.CosineSimilaritycs,eZdZdZdejjdffdd ZZS)CosineSimilaritya8 Computes the cosine similarity between labels and predictions. Note that it is a number between -1 and 1. When it is a negative number between -1 and 0, 0 indicates orthogonality and values closer to -1 indicate greater similarity. The values closer to 1 indicate greater dissimilarity. This makes it usable as a loss function in a setting where you try to maximize the proximity between predictions and targets. If either `y_true` or `y_pred` is a zero vector, cosine similarity will be 0 regardless of the proximity between predictions and targets. `loss = -sum(l2_norm(y_true) * l2_norm(y_pred))` Standalone usage: >>> y_true = [[0., 1.], [1., 1.]] >>> y_pred = [[1., 0.], [1., 1.]] >>> # Using 'auto'/'sum_over_batch_size' reduction type. >>> cosine_loss = tf.keras.losses.CosineSimilarity(axis=1) >>> # l2_norm(y_true) = [[0., 1.], [1./1.414, 1./1.414]] >>> # l2_norm(y_pred) = [[1., 0.], [1./1.414, 1./1.414]] >>> # l2_norm(y_true) . l2_norm(y_pred) = [[0., 0.], [0.5, 0.5]] >>> # loss = mean(sum(l2_norm(y_true) . l2_norm(y_pred), axis=1)) >>> # = -((0. + 0.) + (0.5 + 0.5)) / 2 >>> cosine_loss(y_true, y_pred).numpy() -0.5 >>> # Calling with 'sample_weight'. >>> cosine_loss(y_true, y_pred, sample_weight=[0.8, 0.2]).numpy() -0.0999 >>> # Using 'sum' reduction type. >>> cosine_loss = tf.keras.losses.CosineSimilarity(axis=1, ... reduction=tf.keras.losses.Reduction.SUM) >>> cosine_loss(y_true, y_pred).numpy() -0.999 >>> # Using 'none' reduction type. >>> cosine_loss = tf.keras.losses.CosineSimilarity(axis=1, ... reduction=tf.keras.losses.Reduction.NONE) >>> cosine_loss(y_true, y_pred).numpy() array([-0., -0.999], dtype=float32) Usage with the `compile()` API: ```python model.compile(optimizer='sgd', loss=tf.keras.losses.CosineSimilarity(axis=1)) ``` Args: axis: The axis along which the cosine similarity is computed (the features axis). Defaults to -1. reduction: Type of `tf.keras.losses.Reduction` to apply to loss. Default value is `AUTO`. `AUTO` indicates that the reduction option will be determined by the usage context. For almost all cases this defaults to `SUM_OVER_BATCH_SIZE`. When used with `tf.distribute.Strategy`, outside of built-in training loops such as `tf.keras` `compile` and `fit`, using `AUTO` or `SUM_OVER_BATCH_SIZE` will raise an error. Please see this custom training [tutorial] (https://www.tensorflow.org/tutorials/distribute/custom_training) for more details. name: Optional name for the instance. rVrcstjt|||ddS)N)rrrZ)rBr%r)r"rZrr)r(r#r$r%szCosineSimilarity.__init__) r)r;r<r=r rr9r%rLr#r#)r(r$rs?rcCs>t|tp8t|tr|jtkp8t|dr2|jdkp8|dk}|S)Nr)r\)r{r[rArCr\hasattrr))lossresultr#r#r$is_categorical_crossentropys     rzkeras.losses.serializecCst|S)zSerializes loss function or `Loss` instance. Args: loss: A Keras `Loss` instance or a loss function. Returns: Loss configuration dictionary. )r)rr#r#r$ serialize s rzkeras.losses.deserializecCst|t|ddS)a>Deserializes a serialized loss class/function instance. Args: name: Loss configuration. custom_objects: Optional dictionary mapping names (strings) to custom objects (classes and functions) to be considered during deserialization. Returns: A Keras `Loss` instance or a loss function. z loss function)Zmodule_objectscustom_objectsZprintable_module_name)rglobals)rrr#r#r$ deserialize-s rzkeras.losses.getcCsV|dkr dSt|tr&t|}t|St|tr8t|St|rD|Std|dS)a^Retrieves a Keras loss as a `function`/`Loss` class instance. The `identifier` may be the string name of a loss function or `Loss` class. >>> loss = tf.keras.losses.get("categorical_crossentropy") >>> type(loss) >>> loss = tf.keras.losses.get("CategoricalCrossentropy") >>> type(loss) You can also specify `config` of the loss to this function by passing dict containing `class_name` and `config` as an identifier. Also note that the `class_name` must map to a `Loss` class >>> identifier = {"class_name": "CategoricalCrossentropy", ... "config": {"from_logits": True}} >>> loss = tf.keras.losses.get(identifier) >>> type(loss) Args: identifier: A loss identifier. One of None or string name of a loss function/class or loss configuration dictionary or a loss function or a loss class instance. Returns: A Keras loss as a `function`/ `Loss` class instance. Raises: ValueError: If `identifier` cannot be interpreted. Nz.Could not interpret loss function identifier: )r{strrrHcallabler:) identifierr#r#r$get@s"  rZint32)F)r)FrrV)FrrV)FrV)FrV)FrrV)FrrV)rV)N)lr=r?rZ tensorflow.python.autograph.corerZ tensorflow.python.autograph.implrr-Ztensorflow.python.distributerZtensorflow.python.eagerrZtensorflow.python.frameworkrrrr r Ztensorflow.python.kerasr Ztensorflow.python.keras.utilsr r Z+tensorflow.python.keras.utils.generic_utilsrrZtensorflow.python.opsrrrrZtensorflow.python.ops.lossesrZtensorflow.python.ops.raggedrrrZtensorflow.python.utilrZ tensorflow.python.util.tf_exportrZtensorflow.tools.docsrrrArMrOrQrSrUr[r]r_rarcrergrirkZadd_dispatch_supportrNrZdispatch_for_typesr|rrPrrRrrTrrrbr`rdrnrhr\rr^rrWrrjrfrrZbceZBCEZmseZMSEZmaeZMAEZmapeZMAPEZmsleZMSLEZkldZKLDZkullback_leibler_divergenceZlogcoshrlrrrrZsparse_softmax_cross_entropyZLABEL_DTYPES_FOR_LOSSESr#r#r#r$s>                         )599<<fRK:<:78;B Q " "  &* # )  $"#J    0