o #j+@sddlmZmZddlmZmZmZddlmZddl m Z m Z m Z ddl mZgZGdd d eZGd d d eZGd d d eZGdddeZdS))_C_opspir)core framework unique_name)check_variable_and_dtype)_current_expected_placein_dygraph_mode in_pir_mode) Initializerc,eZdZdZd fdd Zd dd ZZS) NormalInitializera9Implements the Random Normal(Gaussian) distribution initializer Args: loc (float, optional): mean of the normal distribution. Default is 0.0. scale (float, optional): standard deviation of the normal distribution. Default is 1.0. seed (int, optional): random seed. Default is 0. ?rcD|dusJ|dus J|dusJt||_||_||_dSNsuper__init___mean_std_dev_seedselflocscaleseed __class__]/var/www/html/Deteccion_Ine/venv/lib/python3.10/site-packages/paddle/nn/initializer/normal.pyr'     zNormalInitializer.__init__Nc Cs||}t|tjtjfsJt|dgdd|jdkr#|jj|_t r?t }t |j |j|j|j|j|}||dStrVt }t |j |j|j|j|j|}|S|jdd|i|j |j|j|j|jddd d }||_|S) aWInitialize the input tensor with Normal distribution. Args: var(Tensor): Tensor that needs to be initialized. block(Block, optional): The block in which initialization ops should be added. Used in static graph only, default None. Returns: The initialization op. Out)Zuint16Zfloat16Zfloat32Zfloat64Zguassian_randomrNZgaussian_randomF)shapedtypemeanstdrZ use_mkldnnTtypeoutputsattrsZ stop_gradient) _check_block isinstancerBlockrrrprogram random_seedr r rZgaussianr%rrr&_share_underline_tensor_tor append_opop)rvarblockZplaceout_varr4r!r!r"forward0s^     zNormalInitializer.forwardrrrr__name__ __module__ __qualname____doc__rr8 __classcell__r!r!rr"r  rc"eZdZdZdfdd ZZS)Normala`The Random Normal (Gaussian) distribution initializer. Args: mean (float, optional): mean of the normal distribution. Default is 0.0. std (float, optional): standard deviation of the normal distribution. Default is 1.0. name(str, optional): The default value is None. Normally there is no need for user to set this property. For more information, please refer to :ref:`api_guide_Name`. Default: None. Returns: A parameter initialized by Random Normal (Gaussian) distribution. Examples: .. code-block:: python >>> import paddle >>> data = paddle.ones(shape=[3, 1, 2], dtype='float32') >>> weight_attr = paddle.framework.ParamAttr( ... name="linear_weight", ... initializer=paddle.nn.initializer.Normal(mean=0.0, std=2.0)) >>> bias_attr = paddle.framework.ParamAttr( ... name="linear_bias", ... initializer=paddle.nn.initializer.Normal(mean=0.0, std=2.0)) >>> # doctest: +SKIP('name has been used') >>> linear = paddle.nn.Linear(2, 2, weight_attr=weight_attr, bias_attr=bias_attr) >>> print(linear.weight) Parameter containing: Tensor(shape=[2, 2], dtype=float32, place=Place(cpu), stop_gradient=False, [[ 2.1973135 -2.2697184], [-1.9104223 -1.0541488]]) >>> print(linear.bias) Parameter containing: Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=False, [ 0.7885926 -0.74719954]) >>> res = linear(data) >>> print(res) Tensor(shape=[3, 1, 2], dtype=float32, place=Place(cpu), stop_gradient=False, [[[ 1.0754838 -4.071067 ]], [[ 1.0754838 -4.071067 ]], [[ 1.0754838 -4.071067 ]]]) rrNc6|dusJd|dusJdtj||dddSNzmean should not be Nonezstd should not be Noner)rrrrrrr'r(namerr!r"rzNormal.__init__rrNr;r<r=r>rr?r!r!rr"rBrs*rBcr) TruncatedNormalInitializeraRImplements the Random TruncatedNormal(Gaussian) distribution initializer Args: loc (float, optional): Mean of the normal distribution. Default is :math:`0.0`. scale (float, optional): Standard deviation of the normal distribution. Default is :math:`1.0`. seed (int, optional): random seed. Default is 0. rrrcrrrrrr!r"rr#z#TruncatedNormalInitializer.__init__Nc Csx||}t|tjs Jt|tjsJ|jdkr|jj|_|jt j j j t j j j fvrJt j j j}|jtdd|jdg|j|t j j jdd}n|j}|}trt|j|j|j|j|t}|jt j j j t j j j fvr|t||j}||dS||dS|jdd|i|j||j|j|jd d d }|jt j j j t j j j fvr|jd d |id|i|j|jdd||_|S)a_Initialize the input tensor with TruncatedNormal distribution. Args: var(Tensor): Tensor that needs to be initialized. block(Block, optional): The block in which initialization ops should be added. Used in static graph only, default None. Returns: The initialization op r.truncated_gaussian_randomtmpF)rGr%r&r*Z persistableNr$)r%r&r'r(rTr)castX)Zin_dtype out_dtype)r*Zinputsr+r,) r-r.rVariabler/rr0r1r&rZVarDescZVarTypeZFP16ZBF16ZFP32Z create_varrgeneratejoinrGr%Z LOD_TENSORr rrMrrr rOr2r3r4)rr5r6rQr7Zvar_tmpr4r!r!r"r8sv         z"TruncatedNormalInitializer.forwardr9rr:r!r!rr"rKr@rKcrA)TruncatedNormalacThe truncated normal distribution (Gaussian distribution) initializer. Args: mean (float, optional): Mean of the normal distribution. Default is :math:`0.0`. std (float, optional): Standard deviation of the normal distribution. Default is :math:`1.0`. name (str, optional): For details, please refer to :ref:`api_guide_Name`. Generally, no setting is required. Default: None. Returns: A parameter initialized by truncated normal distribution (Gaussian distribution). Examples: .. code-block:: python >>> import paddle >>> data = paddle.ones(shape=[3, 1, 2], dtype='float32') >>> weight_attr = paddle.framework.ParamAttr( ... name="linear_weight", ... initializer=paddle.nn.initializer.TruncatedNormal(mean=0.0, std=2.0)) >>> bias_attr = paddle.framework.ParamAttr( ... name="linear_bias", ... initializer=paddle.nn.initializer.TruncatedNormal(mean=0.0, std=2.0)) >>> # doctest: +SKIP('name has been used') >>> linear = paddle.nn.Linear(2, 2, weight_attr=weight_attr, bias_attr=bias_attr) >>> print(linear.weight) Parameter containing: Tensor(shape=[2, 2], dtype=float32, place=Place(cpu), stop_gradient=False, [[-1.0981836 1.4140984], [ 3.1390522 -2.8266568]]) >>> print(linear.bias) Parameter containing: Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=False, [ -2.1546738 -1.6570673]) >>> res = linear(data) >>> print(res) Tensor(shape=[3, 1, 2], dtype=float32, place=Place(cpu), stop_gradient=False, [[[-0.11380529 -3.0696259 ]], [[-0.11380529 -3.0696259 ]], [[-0.11380529 -3.0696259 ]]]) rrNcrCrDrErFrr!r"r2rHzTruncatedNormal.__init__rIrJr!r!rr"rUs)rUN)ZpaddlerrbaserrrZbase.data_feederrZbase.frameworkr r r Z initializerr __all__rrBrKrUr!r!r!r"s  U1e