[Th"SSKrSSKJr SSKJr SSKJr SSKJr SSKJ r J r SSK J r SSK Jr S /r"S S \5rg) N)Tensor) constraints) Distribution) Independent)ComposeTransform Transform)_sum_rightmost)_sizeTransformedDistributionc*^\rSrSr%Sr0r\\\R4\ S'SU4Sjjr SU4Sjjr \R"SS9S5r\S \4S j5r\R&"54S jr\R&"54S \S \4S jjrSrSrSrSrSrU=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_constraintsc>[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 d/var/www/auris/envauris/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 .r validate_args)rr transformslistall ValueError batch_shape event_shapelenrdomain event_dim forward_shape inverse_shapeexpandr base_distcodomainmaxsuper__init__)selfbase_distributionrr base_shapebase_event_dim transformr$expanded_base_shapebase_batch_shapereinterpreted_batch_ndimstransform_change_in_event_dimr#cutrr __class__s rr+ TransformedDistribution.__init__4s  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)r,r _instancenewshaperr2r6s rr&TransformedDistribution.expandhs(()@)Ljj- ...$//*AOOE*E+ !O3u:DNN4N4N0O#OP--.>?  %s4 )) 5 "00 r8F) is_discretec:UR(dURR$URSRn[ UR 5UR :a7[R"U[ UR 5UR - 5nU$)N) rr'supportr(r!r r#r independent)r,rFs rrFTransformedDistribution.supportwsx>>)) )//"%.. t 7#4#4 4!--T--.1B1BBGr8returnc.URR$r)r' has_rsample)r,s rrK#TransformedDistribution.has_rsamples~~)))r8c[R"5 URRU5nURH nU"U5nM UsSSS5 $!,(df  g=f)z Generates a sample_shape shaped sample or sample_shape shaped batch of samples if the distribution parameters are batched. Samples first from base distribution and applies `transform()` for every transform in the list. N)r;no_gradr'samplerr, sample_shapexr0s rrOTransformedDistribution.samplesE]]_%%l3A!__ aL- __s 8A A&rQcrURRU5nURH nU"U5nM U$)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'rsamplerrPs rrUTransformedDistribution.rsamples4 NN " "< 0I! A)r8c0UR(aURU5 [UR5nSnUn[ UR 5HynUR U5nX%RRURR- - nU[URXd5X%RR- 5- nUnM{ U[URRU5U[URR5- 5-nU$)z Scores the sample by inverting the transform(s) and computing the score using the score of the base distribution and the log abs det jacobian. g)r>_validate_sampler!r r=rinvr"r#r(r log_abs_det_jacobianr'log_prob)r,valuer#r[yr0rRs rr[ TransformedDistribution.log_probs     ! !% ((()  !$//2I a A ))33i6H6H6R6RR RI...q4,,666#HA3n NN # #A & C8R8R4S(S  r8cSnURHnX#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)r,r\rar0s r_monotonize_cdf'TransformedDistribution._monotonize_cdfsM I..(D) dC TQYLs{#c))r8c 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. NrE)rrYr>r'rXcdfrcr,r\r0s rrfTransformedDistribution.cdfsm 2.IMM%(E/    NN + +E 2""5)$$U+ r8cURU5nURRU5nURH nU"U5nM U$)z| Computes the inverse cumulative distribution function using transform(s) and computing the score of the base distribution. )rcr'icdfrrgs rrjTransformedDistribution.icdfsE $$U+##E*Ie$E) r8)r'rr)__name__ __module__ __qualname____firstlineno____doc__rdictstrr Constraint__annotations__r+r&dependent_propertyrFpropertyboolrKr;r<rOr rrUr[rcrfrj__static_attributes__ __classcell__)r6s@rr r sB:rs7 +97F4 % %IlIr8