tjgdZddlZddlZddlmZddlmZmZddlm Z ddl Z ddl m Z gdZ edZ e d Zed Zed Z d:d e d edede jdzde f dZ d:d e dedede jdzde f dZ d:d e deded edede jdzde fdZd e dede fdZd e de fdZ d:dedeezdzdefdZ d;d e d edede jdzde f dZ d;d e dedede jdzde f dZ dd e d+ede jdzde fd,Z$ d>d e d+ede jdzde fd-Z%d e d.edefd/Z& d?d e d ed.edede jdzde f d2Z' d?d e d ed.edede jdzde f d3Z( d@d e d+ede jdzde fd4Z) dAd e d6edede jdzde f d7Z*d8eee fdeee ffd9Z+e+eZ,e+eZ-e+eZ.e+e Z/e+e!Z0e+e$Z1e+e%Z2e+e'Z3e+e(Z4e+e)Z5e+e*Z6dS)BzHThis file contains utilities for initializing neural network parameters.N)Callable)LiteralTypeVar) ParamSpec)Tensor)calculate_gainuniform_normal_ trunc_normal_ constant_ones_zeros_eye_dirac_xavier_uniform_xavier_normal_kaiming_uniform_kaiming_normal_ orthogonal_sparse_uniformnormalconstanteyediracxavier_uniform xavier_normalkaiming_uniformkaiming_normal orthogonalsparse_R_P) linearconv1dconv2dconv3dconv_transpose1dconv_transpose2dconv_transpose3dsigmoidtanhrelu leaky_reluselu)fan_infan_outtensorab generatorreturnctj5||||cdddS#1swxYwYdSNr5)torchno_gradr r2r3r4r5s K/var/www/html/CA-Chatbot/venv/lib/python3.11/site-packages/torch/nn/init.py_no_grad_uniform_r>Es ::q!y99:::::::::::::::::: 9==meanstdctj5||||cdddS#1swxYwYdSr8)r:r;r r2r@rAr5s r=_no_grad_normal_rDLs >>~~dC9~==>>>>>>>>>>>>>>>>>>r?c |jr|Sdtdtfd}||d|zz ks ||d|zzkrtjddt j5|||z |z |||z |z z }|dkr||d }||d } |||| } | |k| | kz} | sn?t j | t j | ||| | } a|| ur| | n?t|t||} d | |z |z dzz} t j |}t j |}|||| | ||||dd | || |}| sn |} |||| t j ||| } |||dd | t j ||| ||}| sn| | |cdddS#1swxYwYdS) Nxr6c`dtj|tjdz zdz S)N?@)matherfsqrt)rFs r=norm_cdfz(_no_grad_trunc_normal_..norm_cdfbs)dhq49S>>1222c99zjmean is more than 2 std from [a, b] in nn.init.trunc_normal_. The distribution of values may be incorrect. stacklevelg333333?cpu)devicer9Tg)is_metafloatwarningswarnr:r; new_tensoritemr anywhere empty_likecopy_maxminr sub_div_pow_mul_log_gt)r2r@rAr3r4r5rMplohiresultmaskmodelog_peak candidates accept_bufpendings r=_no_grad_trunc_normal_rpVs~ :E:e:::: q1s7{q1s7{ 2 2  ;    44 Ha$h#% & &1t8s2B)C)C C s77""1U"3388::B""1U"3388::B^^D#^CCF   4xxzz$V,,44T3)4TT  V## V$$$q#dA,,''Dt s2q88H)&11J)&11J OOAqIO 6 6 6   V $ $ $ OOD ! ! & &s + + 0 0 3 3 8 8 > > C CH M M M ))I)>>CCEEHHTTG;;== % ''1 'BBB"[*fEEFOOD))..s3388;;@@FFKKHUUU#k"++i+@@EEGGJJ:VVG #;;==  V$$$i444444444444444444sMN;;N?N?valctj5||cdddS#1swxYwYdSN)r:r;fill_r2rqs r=_no_grad_fill_rvs !!||C  !!!!!!!!!!!!!!!!!!s 6::ctj5|cdddS#1swxYwYdSrs)r:r;zero_r2s r=_no_grad_zero_rzs} ||~~s 599 nonlinearityparamcgd}||vs|dkrdS|dkrdS|dkrtjdS|dkrw|d }nUt|tst|tst|t r|}nt d |d tjdd|d zzz S|dkr dSt d|)aReturn the recommended gain value for the given nonlinearity function. The values are as follows: ================= ==================================================== nonlinearity gain ================= ==================================================== Linear / Identity :math:`1` Conv{1,2,3}D :math:`1` Sigmoid :math:`1` Tanh :math:`\frac{5}{3}` ReLU :math:`\sqrt{2}` Leaky Relu :math:`\sqrt{\frac{2}{1 + \text{negative\_slope}^2}}` SELU :math:`\frac{3}{4}` ================= ==================================================== .. warning:: In order to implement `Self-Normalizing Neural Networks `__, you should use ``nonlinearity='linear'`` instead of ``nonlinearity='selu'``. This gives the initial weights a variance of ``1 / N``, which is necessary to induce a stable fixed point in the forward pass. In contrast, the default gain for ``SELU`` sacrifices the normalization effect for more stable gradient flow in rectangular layers. Args: nonlinearity: the non-linear function (`nn.functional` name) param: optional parameter for the non-linear function Examples: >>> gain = nn.init.calculate_gain( ... "leaky_relu", 0.2 ... ) # leaky_relu with negative_slope=0.2 )r$r%r&r'r(r)r*r+r,g?r-rIr.N{Gz?znegative_slope z not a valid numberrOr/g?zUnsupported nonlinearity )rJrL isinstanceboolintrU ValueError)r{r| linear_fnsnegative_slopes r=rrsLJz!!\Y%>%>q   w   y~~  % % =!NN5$'' K5#&& K%'' K #NNIuIIIJJ JyNA$5 56777      C\CCDDDrNrHctj|r+tjt|f||||St ||||S)aFill the input Tensor with values drawn from the uniform distribution. :math:`\mathcal{U}(a, b)`. Args: tensor: an n-dimensional `torch.Tensor` a: the lower bound of the uniform distribution b: the upper bound of the uniform distribution generator: the torch Generator to sample from (default: None) Examples: >>> w = torch.empty(3, 5) >>> nn.init.uniform_(w) r<)r: overrideshas_torch_function_variadichandle_torch_functionr r>r<s r=r r s`( 226:: 44 vi!qI5    VQ9 5 55rNctj|r+tjt|f||||St ||||S)aFill the input Tensor with values drawn from the normal distribution. :math:`\mathcal{N}(\text{mean}, \text{std}^2)`. Args: tensor: an n-dimensional `torch.Tensor` mean: the mean of the normal distribution std: the standard deviation of the normal distribution generator: the torch Generator to sample from (default: None) Examples: >>> w = torch.empty(3, 5) >>> nn.init.normal_(w) rC)r:rrrr rDrCs r=r r s`( 226:: 44 fYvDcY5    FD#y 9 99rNrIc,t||||||S)aFill the input Tensor with values drawn from a truncated normal distribution. The values are effectively drawn from the normal distribution :math:`\mathcal{N}(\text{mean}, \text{std}^2)` with values outside :math:`[a, b]` redrawn until they are within the bounds. The method used for generating the random values works best when :math:`a \leq \text{mean} \leq b`. For reduced-precision types (``torch.float16`` and ``torch.bfloat16``), sampling quality depends on the underlying ``normal_()`` and ``uniform_()`` implementations which operate at higher internal precision to avoid quantization artifacts. Args: tensor: an n-dimensional `torch.Tensor` mean: the mean of the normal distribution std: the standard deviation of the normal distribution a: the minimum cutoff value b: the maximum cutoff value generator: the torch Generator to sample from (default: None) Examples: >>> w = torch.empty(3, 5) >>> nn.init.trunc_normal_(w) r9)rp)r2r@rAr3r4r5s r=r r -s B "&$QY O O OOrNctj|r)tjt|f||St ||S)zFill the input Tensor with the value :math:`\text{val}`. Args: tensor: an n-dimensional `torch.Tensor` val: the value to fill the tensor with Examples: >>> w = torch.empty(3, 5) >>> nn.init.constant_(w, 0.3) ru)r:rrrr rvrus r=r r QsX 226:: 44 yS5    &# & &&rNc"t|dS)zFill the input Tensor with the scalar value `1`. Args: tensor: an n-dimensional `torch.Tensor` Examples: >>> w = torch.empty(3, 5) >>> nn.init.ones_(w) rH)rvrys r=r r cs &# & &&rNc t|S)zFill the input Tensor with the scalar value `0`. Args: tensor: an n-dimensional `torch.Tensor` Examples: >>> w = torch.empty(3, 5) >>> nn.init.zeros_(w) )rzrys r=rrps & ! !!rNc|dkrtdtj5tj|j||jddddn #1swxYwY|S)a=Fill the 2-dimensional input `Tensor` with the identity matrix. Preserves the identity of the inputs in `Linear` layers, where as many inputs are preserved as possible. Args: tensor: a 2-dimensional `torch.Tensor` Examples: >>> w = torch.empty(3, 5) >>> nn.init.eye_(w) rO,Only tensors with 2 dimensions are supported)out requires_gradN) ndimensionrr:r;rshaperrys r=rr}saGHHH QQ 61, each group of channels preserves identity Args: tensor: a {3, 4, 5}-dimensional `torch.Tensor` groups (int, optional): number of groups in the conv layer (default: 1) Examples: >>> w = torch.empty(3, 16, 5, 5) >>> nn.init.dirac_(w) >>> w = torch.empty(3, 24, 5, 5) >>> nn.init.dirac_(w, 3) )z5Only tensors with 3, 4, or 5 dimensions are supportedrz!dim 0 must be divisible by groupsr~rrOrN) rrsizerTr_r:r;rxrange)r2r dimensionssizesout_chans_per_grpmin_dimgds r=rrs( ""$$J""PQQQ KKMME Qx&A<=== ~ aF*#U1X..G  v  A7^^  ??PQF10014aQ19LLMM1__  --1 A!+ A!+- --1 A!+ A!+ A!+ -  , MsC8FF #F c&|}|dkrtd|d}|d}d}|dkr|jddD]}||z}||z}||z}||fS)NrOzNFan in and fan out can not be computed for tensor with fewer than 2 dimensionsr~r)dimrrr)r2rnum_input_fmapsnum_output_fmapsreceptive_field_sizesr0r1s r=_calculate_fan_in_and_fan_outrsJA~~ \   kk!nnO{{1~~ zz||aabb! & &A A % 3 3F!55G 7?rNgainct|\}}|tjdt||zz z}tjd|z}t || ||S)aFill the input `Tensor` with values using a Xavier uniform distribution. The method is described in `Understanding the difficulty of training deep feedforward neural networks` - Glorot, X. & Bengio, Y. (2010). The resulting tensor will have values sampled from :math:`\mathcal{U}(-a, a)` where .. math:: a = \text{gain} \times \sqrt{\frac{6}{\text{fan\_in} + \text{fan\_out}}} Also known as Glorot initialization. Args: tensor: an n-dimensional `torch.Tensor` gain: an optional scaling factor generator: the torch Generator to sample from (default: None) Examples: >>> w = torch.empty(3, 5) >>> nn.init.xavier_uniform_(w, gain=nn.init.calculate_gain("relu")) rI@)rrJrLrUr>)r2rr5r0r1rAr3s r=rrsc44F;;OFG 3v'7!8!8899 9C #A VaRI 6 66rNct|\}}|tjdt||zz z}t |d||S)aFill the input `Tensor` with values using a Xavier normal distribution. The method is described in `Understanding the difficulty of training deep feedforward neural networks` - Glorot, X. & Bengio, Y. (2010). The resulting tensor will have values sampled from :math:`\mathcal{N}(0, \text{std}^2)` where .. math:: \text{std} = \text{gain} \times \sqrt{\frac{2}{\text{fan\_in} + \text{fan\_out}}} Also known as Glorot initialization. Args: tensor: an n-dimensional `torch.Tensor` gain: an optional scaling factor generator: the torch Generator to sample from (default: None) Examples: >>> w = torch.empty(3, 5) >>> nn.init.xavier_normal_(w) rIr)rrJrLrUrD)r2rr5r0r1rAs r=rrsO24F;;OFG 3v'7!8!8899 9C FCi 8 88rNrkc|}ddg}||vrtd|d|t|\}}|dkr|n|S)Nr0r1zMode z" not supported, please use one of )lowerrr)r2rk valid_modesr0r1s r=_calculate_correct_fanrsh ::<>> w = torch.empty(3, 5) >>> nn.init.kaiming_uniform_(w, mode="fan_in", nonlinearity="relu") Note: Be aware that ``fan_in`` and ``fan_out`` are calculated assuming that the weight matrix is used in a transposed manner, (i.e., ``x @ w.T`` in ``Linear`` layers, where ``w.shape = [fan_out, fan_in]``). This is important for correct initialization. If you plan to use ``x @ w``, where ``w.shape = [fan_in, fan_out]``, pass in a transposed weight matrix, i.e. ``nn.init.kaiming_uniform_(w.T, ...)``. )r2r3rkr{r5r,Initializing zero-element tensors is a no-oprOrPrr9N)r:rrrrrrVrWrrrJrLr;r ) r2r3rkr{r5fanrrAbounds r=rr*s[V 226::  44  I%5    FL DQRSSSS  . .C , * *D 3 C IcNNS E CCvu BBCCCCCCCCCCCCCCCCCCsC44C8;C8c<d|jvrtjdd|St||}t ||}|t j|z }tj5| d||cdddS#1swxYwYdS)aFill the input `Tensor` with values using a Kaiming normal distribution. The method is described in `Delving deep into rectifiers: Surpassing human-level performance on ImageNet classification` - He, K. et al. (2015). The resulting tensor will have values sampled from :math:`\mathcal{N}(0, \text{std}^2)` where .. math:: \text{std} = \frac{\text{gain}}{\sqrt{\text{fan\_mode}}} Also known as He initialization. Args: tensor: an n-dimensional `torch.Tensor` a: the negative slope of the rectifier used after this layer (only used with ``'leaky_relu'``) mode: either ``'fan_in'`` (default) or ``'fan_out'``. Choosing ``'fan_in'`` preserves the magnitude of the variance of the weights in the forward pass. Choosing ``'fan_out'`` preserves the magnitudes in the backwards pass. nonlinearity: the non-linear function (`nn.functional` name), recommended to use only with ``'relu'`` or ``'leaky_relu'`` (default). generator: the torch Generator to sample from (default: None) Examples: >>> w = torch.empty(3, 5) >>> nn.init.kaiming_normal_(w, mode="fan_out", nonlinearity="relu") Note: Be aware that ``fan_in`` and ``fan_out`` are calculated assuming that the weight matrix is used in a transposed manner, (i.e., ``x @ w.T`` in ``Linear`` layers, where ``w.shape = [fan_out, fan_in]``). This is important for correct initialization. If you plan to use ``x @ w``, where ``w.shape = [fan_in, fan_out]``, pass in a transposed weight matrix, i.e. ``nn.init.kaiming_normal_(w.T, ...)``. rrrOrPr9N) rrVrWrrrJrLr:r;r )r2r3rkr{r5rrrAs r=rrksV FL DQRSSSS  . .C , * *D 3 C ;;~~a ~::;;;;;;;;;;;;;;;;;;s,BBBc |dkrtd|dks|jr|S|d}||z}|||fdd|}||kr|tj |\}}tj |d}| } || z}||kr|tj 5|||||dddn #1swxYwY|S)aFill the input `Tensor` with a (semi) orthogonal matrix. Described in `Exact solutions to the nonlinear dynamics of learning in deep linear neural networks` - Saxe, A. et al. (2013). The input tensor must have at least 2 dimensions, and for tensors with more than 2 dimensions the trailing dimensions are flattened. Args: tensor: an n-dimensional `torch.Tensor`, where :math:`n \geq 2` gain: optional scaling factor generator: the torch Generator to sample from (default: None) Examples: >>> # xdoctest: +REQUIRES(env:TORCH_DOCTEST_LAPACK) >>> w = torch.empty(3, 5) >>> nn.init.orthogonal_(w) rOz4Only tensors with 2 or more dimensions are supportedrr~r9N)rrnumelrTr new_emptyr t_r:linalgqrdiagsignr;view_asr]rc) r2rr5rowscols flattenedqrrphs r=rrs,QOPPP ||~~fn ;;q>>D <<>>T !D  $..66q!y6QQI d{{  FF Frsparsityc|dkrtd|jr|S|j\}}t j||z}t j5|d||t|D]'}t j |}|d|} d|| |f<( dddn #1swxYwY|S)aFill the 2D input `Tensor` as a sparse matrix. The non-zero elements will be drawn from the normal distribution :math:`\mathcal{N}(0, 0.01)`, as described in `Deep learning via Hessian-free optimization` - Martens, J. (2010). Args: tensor: an n-dimensional `torch.Tensor` sparsity: The fraction of elements in each column to be set to zero std: the standard deviation of the normal distribution used to generate the non-zero values generator: the torch Generator to sample from (default: None) Examples: >>> w = torch.empty(3, 5) >>> nn.init.sparse_(w, sparsity=0.1) rOrrr9N) rrrTrrJceilr:r;r rrandperm) r2rrAr5rr num_zeroscol_idx row_indices zero_indicess r=rrs .aGHHH ~ JD$ (T/**I ..q#333T{{ . .G...K&z z2L,-F<( ) ) ................ Ms%ACC Cmethcjdddtjdtjdtffd }dddd |_|_|S) Nargskwargsr6cZtjdddtd|i|S)Nz `nn.init.z)` is now deprecated in favor of `nn.init.z`.rOrP)rVrW FutureWarning)rrrnew_nameold_names r=deprecated_initz(_make_deprecate..deprecated_initsO W W W8 W W W     tT$V$$$rNz z_(...) .. warning:: This method is now deprecated in favor of :func:`torch.nn.init.z"`. See :func:`~torch.nn.init.z` for details.)__name__r#rrr"__doc__)rrrrs` @@r=_make_deprecaters}H}H%rw%")%%%%%%%%%: ::IQ :: ( :::O (O rNrs)rrHN)rrHrrIN)r~)rHN)rr0r.N)r~N)rN)7rrJrVcollections.abcrtypingrrtyping_extensionsrr:r__all__r"r#_NonlinearityType_FanModerU Generatorr>rDrprvrzrrr r r r r rrrtuplerrrrrrrrrrrrrrrrrrr r!rNr=rsNN $$$$$$########''''''    > WT]]Yt__    & 'MQ:: ::!&:38?T3I: ::::)- >> > > >% >  >>>> )- JJ J J J J  J % J JJJJZ!6!!&!!!! 6f BFGEGE#GE,/%K$,>GE GEGEGEGEX(, 66 6 6 6% 6  6666:(, :: : : :% :  :::::(, !P!P !P !P !P !P  !P % !P !P!P!P!PH'f'5'V''''$ '& 'V ' ' ' ' "6 "f " " " "F*556535v5555p&U38_.(,77 7 7%7 7777F(,99 9 9%9 9999>3633c3333&2(, >C>C >C >C >C$ >C % >C  >C>C>C>CF&2(, 2;2; 2; 2; 2;$ 2; % 2;  2;2;2;2;n(,00 0 0%0 0000l(, && && &% &  &&&&T(2r6*xB/?. /( # #  ! ! ?9 % %od 11// !/"233 11 _[ ) )  ! !rN