9i"SSKJrJr SSKrSSKJr SSKJr SSKJr SSK J r SSK J r J r SSKJr SS KJr S /r"S S \5rg) )OptionalUnionN)Tensor) constraints) Distribution) Independent)ComposeTransform Transform)_sum_rightmost)_sizeTransformedDistributionc V^\rSrSr%Sr0r\\\R4\ S'SS\ S\ \ \\ 4S\\SS4U4S jjjrSU4S jjr\R&"S S 9S 5r\S\4Sj5r\R0"54Sjr\R0"54S\S\4SjjrSrSrSrSr Sr!U=r"$)r aI Extension of the Distribution class, which applies a sequence of Transforms to a base distribution. Let f be the composition of transforms applied:: X ~ BaseDistribution Y = f(X) ~ TransformedDistribution(BaseDistribution, f) log p(Y) = log p(X) + log |det (dX/dY)| Note that the ``.event_shape`` of a :class:`TransformedDistribution` is the maximum shape of its base distribution and its transforms, since transforms can introduce correlations among events. An example for the usage of :class:`TransformedDistribution` would be:: # Building a Logistic Distribution # X ~ Uniform(0, 1) # f = a + b * logit(X) # Y ~ f(X) ~ Logistic(a, b) base_distribution = Uniform(0, 1) transforms = [SigmoidTransform().inv, AffineTransform(loc=a, scale=b)] logistic = TransformedDistribution(base_distribution, transforms) For more examples, please look at the implementations of :class:`~torch.distributions.gumbel.Gumbel`, :class:`~torch.distributions.half_cauchy.HalfCauchy`, :class:`~torch.distributions.half_normal.HalfNormal`, :class:`~torch.distributions.log_normal.LogNormal`, :class:`~torch.distributions.pareto.Pareto`, :class:`~torch.distributions.weibull.Weibull`, :class:`~torch.distributions.relaxed_bernoulli.RelaxedBernoulli` and :class:`~torch.distributions.relaxed_categorical.RelaxedOneHotCategorical` arg_constraintsNbase_distribution transforms validate_argsreturnc>[U[5(a U/UlOL[U[5(a)[ SU55(d [ S5eX lO[ SU35eUR UR-n[UR5n[UR5n[U5URR:a&[ SURRSUS35eURU5nURU5nXH:wa"US[U5U- n URU 5nURRU- n U S:a [X5nXlUR"RURR- n [%UR"RX[-5n [U5U :de[U5U - n USU nX}Sn[&TU]QXUS9 g) Nc3B# UHn[U[5v M g7fN) isinstancer ).0ts l/var/www/html/land-doc-ocr/venv/lib/python3.13/site-packages/torch/distributions/transformed_distribution.py 3TransformedDistribution.__init__..@sDAz!Y//sz6transforms must be a Transform or a list of Transformsz0transforms must be a Transform or list, but was z9base_distribution needs to have shape with size at least z , but got .rr)rr rlistall ValueError batch_shape event_shapelenr domain event_dim forward_shape inverse_shapeexpandr base_distcodomainmaxsuper__init__)selfrrr base_shapebase_event_dim transformr(expanded_base_shapebase_batch_shapereinterpreted_batch_ndimstransform_change_in_event_dimr'cutr#r$ __class__s rr/ TransformedDistribution.__init__5s  j) , ,DO D ) )DDDD L)OB:,O  '225F5R5RR .::;$T__5 z?Y--77 7KIL\L\LfLfKggqr|q}}~ "// ; '55mD  ,2;#)*^;  !2 8 89I J $-$4$4$>$>$O! $q ( +!! +    ( (9+;+;+E+E E &    ( (  : =!Y...- 9,#DS) #D)  Oc>UR[U5n[R"U5nXR-n[ UR 5HnURU5nM US[U5[URR5- nURRU5Ul UR Ul[[U]3XRSS9 URUl U$)NFr)_get_checked_instancer torchSizer$reversedrr)r%r+r*r.r/_validate_args)r0r# _instancenewshaperr5r9s rr*TransformedDistribution.expandns(()@)Ljj- ...$//*AOOE*E+ !O3u:DNN4N4N0O#OP--.>?  %s4 )) 5 "00 r;F) is_discretec:UR(dURR$URSRn[ UR 5UR :a7[R"U[ UR 5UR - 5nU$)N) rr+supportr,r%r$r'r independent)r0rIs rrITransformedDistribution.support}sx>>)) )//"%.. t 7#4#4 4!--T--.1B1BBGr;c.URR$r)r+ has_rsample)r0s rrM#TransformedDistribution.has_rsamples~~)))r;c[R"5 URRU5nURH nU"U5nM UsSSS5 $!,(df  g=f)z Generates a sample_shape shaped sample or sample_shape shaped batch of samples if the distribution parameters are batched. Samples first from base distribution and applies `transform()` for every transform in the list. N)r>no_gradr+samplerr0 sample_shapexr3s rrQTransformedDistribution.samplesE]]_%%l3A!__ aL- __s 8A A&rScrURRU5nURH nU"U5nM U$)z Generates a sample_shape shaped reparameterized sample or sample_shape shaped batch of reparameterized samples if the distribution parameters are batched. Samples first from base distribution and applies `transform()` for every transform in the list. )r+rsamplerrRs rrWTransformedDistribution.rsamples4 NN " "< 0I! A)r;c0UR(aURU5 [UR5nSnUn[ UR 5HynUR U5nX%RRURR- - nU[URXd5X%RR- 5- nUnM{ U[URRU5U[URR5- 5-nU$)z Scores the sample by inverting the transform(s) and computing the score using the score of the base distribution and the log abs det jacobian. g)rA_validate_sampler%r$r@rinvr&r'r,r log_abs_det_jacobianr+log_prob)r0valuer'r]yr3rTs rr] TransformedDistribution.log_probs     ! !% ((() ), !$//2I a A ))33i6H6H6R6RR RI...q4,,666#HA3n NN # #A & C8R8R4S(S  r;cSnURHnX#R-nM [U[5(aUS:XaU$X!S- -S-$)z] This conditionally flips ``value -> 1-value`` to ensure :meth:`cdf` is monotone increasing. g?)rsignrint)r0r^rcr3s r_monotonize_cdf'TransformedDistribution._monotonize_cdfsM I..(D) dC TQYLs{#c))r;c URSSS2HnURU5nM UR(aURR U5 URR U5nUR U5nU$)z Computes the cumulative distribution function by inverting the transform(s) and computing the score of the base distribution. NrH)rr[rAr+rZcdfrer0r^r3s rrhTransformedDistribution.cdfsm 2.IMM%(E/    NN + +E 2""5)$$U+ r;cURU5nURRU5nURH nU"U5nM U$)z| Computes the inverse cumulative distribution function using transform(s) and computing the score of the base distribution. )rer+icdfrris rrlTransformedDistribution.icdfsE $$U+##E*Ie$E) r;)r+rr)#__name__ __module__ __qualname____firstlineno____doc__rdictstrr Constraint__annotations__rrr r rboolr/r*dependent_propertyrIpropertyrMr>r?rQr rrWr]rerhrl__static_attributes__ __classcell__)r9s@rr r sB:rtorch.distributionsr torch.distributions.distributionrtorch.distributions.independentrtorch.distributions.transformsr r torch.distributions.utilsr torch.typesr __all__r r;rrs8" +97F4 % %NlNr;