ёi8SSKJr SSKrSSKJrJr SSKJrJrJr SSK J r SSK J r J r Jr SS KJr SS KJr /r"S S \5r"S S\5r"SS\5r"SS\5rg)) annotationsN)_C_opspir)core framework unique_name)check_variable_and_dtype)_current_expected_placein_dygraph_mode in_pir_mode) Initializer)lazy_init_helperc^^\rSrSrSrSSU4SjjjrSS SjjrSrU=r$) NormalInitializer!a-Implements the Random Normal(Gaussian) distribution initializer Args: loc (float|complex, optional): mean of the normal distribution. Default is 0.0. scale (float, optional): standard deviation of the normal distribution. Default is 1.0. seed (int, optional): random seed. Default is 0. c>UceUceUce[TU]5 XlX lX0l[ UR[ 5(aURRURR:wa9[SURRSURR35eURRUlgg)NzVif mean is a complex number, its real part should equal imag part, but got real part: z != imag part: ) super__init___mean_std_dev_seed isinstancecomplexrealimag ValueError)selflocscaleseed __class__s \/var/www/html/banglarbhumi/venv/lib/python3.13/site-packages/paddle/nn/initializer/normal.pyrNormalInitializer.__init__+s       djj' * *zz$**//1 **.**//):/$**//IZ\DJ +c [U[R5(aUR5(aS5eUR U5n[U[R [ R 45(de[US/SQS5 URS:XaURRUl[5(ai[5n[R"URUR UR"URUR$U5nUR'U5 g[)5(aY[5n[R"URUR UR"URUR$U5nU$UR+SSU0URUR$UR UR"URS.SS 9nXQlU$) a$Initialize the input tensor with Normal distribution. Args: var(Tensor): Tensor that needs to be initialized. block(Block|None, optional): The block in which initialization ops should be added. Used in static graph only, default None. Returns: The initialization op. zCCurrently, normal initializer not support lazy init for dist param.Out)uint16float16float32float64 complex64 complex128gaussian_randomrN)shapedtypemeanstdr"Ttypeoutputsattrs stop_gradient)rrEagerParamBaseis_dist _check_blockBlockrr rprogram random_seedr r rgaussianr0rrr1_share_underline_tensor_tor append_opop)rvarblockplaceout_varrBs r$forwardNormalInitializer.forward=s sI44 5 5#++-- Q P Q !!%(%)//399!=>>>>      ::?22DJ   +-Eoo      G  . .s 3 ]]+-Eoo      GN&  YY YY JJ== JJ #! BFIr&)rrr)?r)r floatr!rKr"intreturnNoneNrCz paddle.TensorrDzpir.Block | NonerMzpaddle.Tensor | None __name__ __module__ __qualname____firstlineno____doc__rrG__static_attributes__ __classcell__r#s@r$rr!sfAB))',):=) ))&=AJ J)9J JJr&rcD^\rSrSrSrSSU4SjjjrSrU=r$)NormalaThe Random Normal (Gaussian) distribution initializer. Args: mean (float|complex, 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|None, 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 ]]]) cJ>UcS5eUcS5e[TU]XSS9 g)Nmean should not be Nonestd should not be Noner)r r!r"rr)rr2r3namer#s r$rNormal.__init__s9:!::8 88 T15r&)rIrJN)r2rKr3rKra str | NonerMrNrRrSrTrUrVrrWrXrYs@r$r[r[s:(VGK66&+69C6 66r&r[cn^\rSrSrSrSSU4SjjjrSS SjjrSrU=r$) TruncatedNormalInitializeraTImplements the Random TruncatedNormal(Gaussian) distribution initializer Note: It is better to set `a <= mean <= b`. If `mean < a - 2*std` or `mean > b + 2*std`, the distribution of values may be incorrect. Args: loc (float, optional): Mean of the normal distribution. Default is :math:`0.0`. scale (float, optional): Standard deviation of the normal distribution. Default is :math:`1.0`. seed (int, optional): random seed. Default is 0. a (float, optional): The minimum cutoff value. Default is -2.0. b (float, optional): The maximum cutoff value. Default is 2.0. c>UceUceUceUceUce[TU]5 XlX lX0lX@lXPlgrO)rrrrr_a_b)rr r!r"abr#s r$r#TruncatedNormalInitializer.__init__s_   }}}}    r&c $ URU5n[5R(aN[R[ R RR[RR4nOM[R[ R R[ R RR4n[X5(de[U[R[ R45(deURS:XaURR UlUR"[R$R&R([R$R&R*4;a[R$R&R,nUR/[0R2"SR5SUR6S/55UR8U[R$R&R:SS9nOUR"nUn[=5(a[>R@"UR8URBURDURURFURHU[K55nUR"[R$R&R([R$R&R*4;a2[>RL"XQR"5nUROU5 gUROU5 g[Q5(a[>R@"UR8URBURDURURFURHU[K55nUR"[R$R&R([R$R&R*4;a1[>RL"XQR"5nUROU5 U$URSSSU0UR8UURBURDURURFURHS .S S 9nUR"[R$R&R([R$R&R*4;a-URSS S U0SU0UR"UR"S.S9 Xql*U$)a,Initialize the input tensor with TruncatedNormal distribution. Args: var(Tensor): Tensor that needs to be initialized. block(Block|None, 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)rar0r1r5 persistableNr()r0r1r2r3r"rlrmTr4castX)in_dtype out_dtype)r5inputsr6r7)+r;rstaterVariablepaddlerr ParameterMetaeagerTensorValuerr<rr=r>r1VarDescVarTypeFP16BF16FP32 create_varr generatejoinrar0 DENSE_TENSORr rrqrrrjrkr rtr@r rArB)rrCrDexpectedrwrFvar_tmprBs r$rG"TruncatedNormalInitializer.forwards!!%(   # #"" -- !!H""    --H #((((%)//399!=>>>> ::?22DJ 99--22DLL4H4H4M4MN N ,,11I&& ))HH9388UKLii\\))66!'G IG   66    ') Gyy $$)) $$))!++gyy922372237 ]]66    ') Gyy $$)) $$))!++gyy92237N0( YY& JJ== JJ#! Byy $$)) $$))>"CL'.}}399M   FIr&)rjrkrrr)rIrJr@) r rKr!rKr"rLrlrKrmrKrMrNrOrPrQrYs@r$rgrgs "        *=At t)9t ttr&rgcT^\rSrSrSrSSU4SjjjrSrU=r$)TruncatedNormaliXaThe truncated normal distribution (Gaussian distribution) initializer. Note: It is better to set `a <= mean <= b`. If `mean < a - 2*std` or `mean > b + 2*std`, the distribution of values may be incorrect. 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`. a (float, optional): The minimum cutoff value. Default is -2.0. b (float, optional): The maximum cutoff value. Default is 2.0. name (str|None, 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 ]]]) ct>UcS5eUcS5eUcS5eUcS5e[TU]XSX4S9 g)Nr^r_za should not be Nonezb should not be Noner)r r!r"rlrmr`)rr2r3rlrmrar#s r$rTruncatedNormal.__init__s[:!::8 88}444}}444} T1?r&rc)rIrJrrN) r2rKr3rKrlrKrmrKrardrMrNrerYs@r$rrXsf-b @ @ @  @  @  @  @ @r&r) __future__rr{rrbaserrr base.data_feederr base.frameworkr r r initializerr lazy_initr__all__rr[rgrrcr&r$rsk# 008 %' f fR06 06fXXv<@0<@r&