EiSh BSrSSKrSSKrSSKrSSKrSSKrSSKrSSKJrJ r SSK J r SSK J r SSKJr SSKJrJrJrJr SSKrSSKJr SSKrSSKrSSKJr SS KJr SS KJ r SS K!J"r"J#r#J$r$J%r%J&r& SS K'J(r( S SK)J*r*J+r+J,r, S SK-J.r. S SK/J0r0J1r1J2r2J3r3J4r4 S SK5J6r6J7r7 S SK8J9r9J:r: S SK;Jr>J?r?J@r@JArAJBrBJCrCJDrDJErEJFrFJGrGJHrH SSKIJJrJJKrKJLrL SSKMJNrNJOrO SSKPJQrQJRrR SSKSJTrT SSKUJVrV SSKWrX\(a SSKZJ[r[ SSK\J]r] SSK^J_r_J`r` SSKaJbrb SSKcJdrd \R"\f5rg\R\R\R\R\R\R\R\RS .rp\R\R\R\RS!.rq0\pE\qErr\sR\pR55rv\sR\qR55rwS"\xS#\y4S$jrz\RRR\RR-r"S%S&\L5r"S'S(\L5r"S)S*\5r"S+S,\5r"S-S.\5r"S/S0\V5r"S1S2\L5r"S3S4\L5rg!\Ya SrXGNf=f)5a This module contains variable tracker classes for handling tensors and tensor-related operations in Dynamo. The main class is TensorVariable which represents torch.Tensor inputs and intermediate values in the FX graph. It handles tensor operations, method calls, and maintains metadata about tensor properties like dtype, device, etc. Other key classes include: - SymNodeVariable: Represents symbolic scalars (int/float/bool) used for size computation and unspecialized values - NumpyNdarrayVariable: Handles numpy array interop through torch._numpy - UnspecializedPythonVariable: Represents unspecialized Python numeric values as 1-element tensors - TensorSubclassVariable: Handles tensor subclasses with __torch_function__ overrides - UntypedStorageVariable: Represents tensor storage objects - DataPtrVariable: Handles tensor data pointer operations These classes work together to track tensor operations and properties during Dynamo's tracing process. N)IterableSequence) nullcontext)chain)NoneType)AnyNoReturnOptional TYPE_CHECKING)compiled_autograd)is_opaque_reference_type) is_sparse_any) guard_scalarGuardOnDataDependentSymNodehas_free_symbols is_symbolicSymTypes)is_traceable_wrapper_subclass)configgraph_break_hints variables) trace_wrapped)TorchRuntimeError unimplemented$UnknownPropertiesDuringBackwardTrace UserError UserErrorType)_ApplyBackwardHookcall_hook_from_backward_state) GuardBuilder install_guard) AttrSource) fqnget_custom_getattrget_fake_valueget_real_value guard_if_dynobject_has_getattributeproductproxy_args_kwargsraise_args_mismatchset_example_valuetensortype_to_dtype)AttributeMutationNewValueMutationNewVariableTracker)CONSTANT_VARIABLE_NONEConstantVariable)ListIteratorVariable SizeVariable)TorchScriptObjectVariable)UserDefinedClassVariable) PyCodegen) OutputGraph)InstructionTranslatorInstructionTranslatorBase)UserFunctionVariableTensorWithTFOverrideVariable)>=z<===!=isis not)rDrErBrCvaluereturnc`[[U5=(a [RRR U5(+=(a_ [ US5=(aL [UR[R5=(a! [URURS55$)N__self__) boolcallabletorch_dynamoutilsr)hasattr isinstancerITensorgetattr__name__rFs [/home/mande/repo/quber/.venv/lib/python3.13/site-packages/torch/_dynamo/variables/tensor.pyis_bound_tensor_methodrVsy  : ##;;EB B : E: & : u~~u|| 4 : ENNENND 9  c# ^\rSrSrSrSSSSSSS S S S S SSSS1\R krS\R4Sjr SSSSS.S\RRS\RS\RS\RS\S \S \S \S\S\S\S\\S4S-S \\S4S-S \S-S\S-S\SS4"U4SjjjrSSSS\S-SS4SjjrS\S-4SjrSSS \S-S!\SS4S"jrS\4S#jrS\RR4S$jrS\4S%jrS\4S&jr\S'\RS\\\44S(j5rSSS)\S\4S*jr SSS\4S+jr!SSS\S-4S,jr"SSS\S-4S-jr#SSS\S-4S.jr$SSS\%S-4S/jr&SSS\4S0jr'SSS\%S-4S1jr(SSS\%S-4S2jr)SSS\%S-4S3jr*SSS\%S-4S4jr+SSS\,4S5jr-SSS\4S6jr.SSS\%S-4S7jr/SSS\4S8jr0SSS)\S\%4S9jr1SSS)\S\4S:jr2SSS\4S;jr3SSS\4S<jr4SSSS=\5\S-S\6\4S>jjr7SSS?S@SA\SB\5\SC\\\4S\4 SDjr8S\4SEjr9\:S\\S44SFj5r;S\6\4SGjrSSSI\S\S\S-4SLjr?SSSI\S\S\S-4SMjr@SS)\SN\S-S\S-4SOjjrASSS\S-4SPjrB\BrCSSS\S-4SQjrD\DrESSS\%S-4SRjrFSSS\%S-4SSjrGSSS\%S-4STjrHSSSSU\S-S\%S-4SVjjrISSSS\S-SX\S\S\S-4 SYjjrJSSSZ\SS[4S\jrKSSS\S-4S]jrLSSS\4S^jrMSWS_.SSS`\\-SSa4SbjjrNSSS\4ScjrOSSd\P\Se\S\Q\6\4SfjjrRSSSSg\Q\Sh\Q\Si\Q\Sj\Q\S\Q\4 SkjjrSSSSI\S\SSl4SmjrTSSSI\S\SS4SnjrUSSSI\S\S\4SojrV\\WRSSpj55rYSSS\4SqjrZSSS\[4Srjr\SSs.SSSt\Su\S'\S-S\S-4 Svjjr]SSSw\S'\S\4Sxjr^SSSI\S\S\,4Syjr_SSSI\S\S\,4Szjr`SSSI\S\S\,4S{jraSSSI\S\S\,4S|jrbSSSI\S\SS4S}jrcSS~.SSS\S\S-S\S-4SjjrdSSs.SSStSSuSS'\S-S\S-4 SjjreSSS\S\4SjrfSSSI\S\S\4SjrgSSSI\S\S\4SjrhSSSI\S\S\4SjriSSSI\S\S\4SjrjSSS)\S\S\4SjrkSSSS \\-S\4SjjrlS\,4SjrmSSSI\S\S\S-4SjrnSSjroS)\SS4SjrpS\4SjrqS\4SjrrS\sS\4SjrtSruU=rv$)TensorVariablez=A torch.Tensor input or an intermediate value in the FX graphproxydtypedevicelayoutndimsizestride requires_grad is_quantized is_contiguous is_nested is_sparse class_typespecialized_value _is_name_setrGcj[URRURR5$)z Get the actual value represented by this variable if computation is run using the user-provided inputs. NOTE: this runs actual tensor computation and may be slow and memory-intensive. )r'r[nodetracerselfs rUr'TensorVariable.get_real_values#djjootzz/@/@AArWN)_sizerardri has_grad_fnrp.kwargsc &>[TU]"S0UD6 XlX lX0lX@lXPlXlXlX`l Xl Xl Xpl Xl XlXlUc#URR R"S:HnXlg)N placeholder)super__init__r[r\r]r^r_rprarbrcrdrerfrgrqrkopri)rnr[r\r]r^r_rbrercrfrgrqrprardrirr __class__s rUrwTensorVariable.__init__s( "6"       *(*""$&  ::??-->L".rWtxr; target_clscSSKJnJn Uc [U5nURR R RS5nU"X!XT"U55nUR5Hupx[XU5 M g)Nr/)get_specialized_propsinfer_subclass_type example_value) builderr~rtyper[rkmetagetitemssetattr) rnr{r|r~rrspecialized_propskvs rUsynchronize_attributes%TensorVariable.synchronize_attributessn H  dJ ,,00A 1 M+>}+M &++-DA DQ .rWcURRRRS5nUb UR$S$)zFGet the current version of self's fake tensor, or None if unavailable.rN)r[rkrr_version)rn self_fakes rU_get_fake_version TensorVariable._get_fake_versions6JJOO((,,_= %.%:y!!DDrWversion_beforehas_tensor_argcrUR5nUb$Ub XB:aU(aURU5 ggggg)zs Sync attributes if self was mutated by an inplace operation. See Note [Inplace ops and VariableTracker metadata] N)rr)rnr{rr version_afters rU_sync_if_inplace_mutation(TensorVariable._sync_if_inplace_mutationsI..0  &).  ' ' +/* 'rWcZ[URRRS5$Nr)reprr[rkrrms rU debug_reprTensorVariable.debug_reprs DJJOO((9::rWcUR$Nr[rms rUas_proxyTensorVariable.as_proxy zzrWcUR$r)rgrms rU python_typeTensorVariable.python_type s rWcgNTrurms rU is_tensorTensorVariable.is_tensor rWrFc &^TRTRTR[TR5TR TR TRTR[T5S. nTRSLUS'[T5(a5[T5(d%[STR555US'U$[T5(d[STR555US'[TR!55US'["R$R&R)T5(aSUS'U$[U4S j["R*R,55US'U$![a SUS'GN f=f) N) r\r]r^r_rbrercrfrgrqFc3\# UH"n[U5(a [U5OUv M$ g7frrint.0ss rU ,TensorVariable.specialize..$s$#9EA+a..Aa/*,rpc3\# UH"n[U5(a [U5OUv M$ g7frrrs rUrr.s*#&A&a..Aa/%rrardc3T># UHnTRUS9(dMUv M g7f) memory_formatN)rd)rxrFs rUrr:s+/@***;A@s( ()r\r]r^rr_rbrercrfrgrad_fn Exceptionrrtupler`rarL_C _functorchis_batchedtensor _prims_common_memory_formats)rFpropss` rU specializeTensorVariable.specializesp[[llll O"00!..u+ !  )#(==##E'N6 1"%((## #E'N $ELLN3E(Oxx""33E::*.o& *//"00@@/*o&  C )$)E-  )s1E==FFnamec8URRRSnUR(Gd [ U5(aUR 5upE[ UR5U5n[ X25nX$;a.[U[R5(deSSK J n U"XUS9$[[U55(aT[RR R#UR$R&U5n [(R*"Xi5$[-U5(d[.R0"X5$UR(aURR35(d[4eUR$R6UR$R8S.n [;URR<U 5n U c[4e[AU 5(a[4e[CU 5(a[4e[ X5n [EURU5n[GU 5(aSSK$J%n U"XU[U 5S9$[MURO[PRR55 [.R0"XU5$![>a n [4U eSn A ff=f)Nrr/ wrap_fx_proxy)r{r[rLGGetAttrVariable)sourcepy_type)*r[rkrrr__tensor_flatten__rRrrPrLrQrrr r_libraryfake_class_registrymaybe_to_fake_objoutput fake_moder7createrKr2buildsubguards_allowedNotImplementedError local_scope global_scopeevalrrr)r%r#rVmiscrr" make_guardr!HASATTR)rnr{rfake_valattrs_ctxr[rrfake_script_objscope_input_associated_real_valueexc real_value attr_sourcers rUdynamic_getattrTensorVariable.dynamic_getattrAs::??''8 {{{>>>2$}UU)$}*=>>"'.."D"D"V"VII''#177OOm,,&,,R??  = = ? ?% %ii++"))2H2HI / ,0 0@0@%+H ( ( /% % "#? @ @% % : ; ;% %9@  d3  "* - - -";Z8H  k,,\-A-ABC$$R[AA7 /%3 . /s' J J JJcURb [R"UR5$URUS/05$)Ndim)r_r4r call_methodrnr{s rUmethod_attr_ndimTensorVariable.method_attr_ndims8 99 #**4995 5##Br26 6rWc^URb [R"UR5$gr)r\r4rrs rUmethod_attr_dtype TensorVariable.method_attr_dtypes$ :: !#**4::6 6rWc^URb [R"UR5$gr)r]r4rrs rUmethod_attr_device!TensorVariable.method_attr_device$ ;; "#**4;;7 7rWc^URb [R"UR5$gr)r^r4rrs rUmethod_attr_layout!TensorVariable.method_attr_layoutrrWcxURb-[R"URRS:H5$g)Ncuda)r]r4rrrs rUmethod_attr_is_cuda"TensorVariable.method_attr_is_cudas1 ;; "#**4;;+;+;v+EF FrWcUR5(aDURVs/sH"n[RR U5PM$ nn[ U5$UR US/05$s snfNr`) valid_sizer`rr4rr6r)rnr{rsizess rUmethod_attr_shape TensorVariable.method_attr_shapesi ??  >Bii,>G **11!4i , & &##BB7 7 ,s)A.c^URb [R"UR5$gr)rbr4rrs rUmethod_attr_requires_grad(TensorVariable.method_attr_requires_grads*    )#**4+=+=> >rWc^URb [R"UR5$gr)rcr4rrs rUmethod_attr_is_quantized'TensorVariable.method_attr_is_quantizeds*    (#**4+<+<= =rWc^URb [R"UR5$gr)rfr4rrs rUmethod_attr_is_sparse$TensorVariable.method_attr_is_sparse& >> %#**4>>: :rWc^URb [R"UR5$gr)rer4rrs rUmethod_attr_is_nested$TensorVariable.method_attr_is_nestedr rWc$[SSUS3S/S9 g)Nz'Tensor.retain_grad() with AOTDispatcher var_getattr z retain_gradz8`Tensor.retain_grad()` does not work with AOTDispatcher.gb_typecontext explanationhintsrrs rUmethod_attr_retain_grad&TensorVariable.method_attr_retain_grads="4& 5R  rWc[R"[RRR 5R X/05$r)rTorchInGraphFunctionVariablerLr _autograd_get_data_attr call_functionrs rUmethod_attr_dataTensorVariable.method_attr_datas655 HH   - - -FB ' (rWcfUR(a[SSUS3S/S9 g[R$)NzTensor with grad_fn()rz grad_fnz@Dynamo does not support tracing tensors with a grad_fn directly.r)rqrrr3rs rUmethod_attr_grad_fn"TensorVariable.method_attr_grad_fns8    /&tfH5^  33 3rWc\SSKJn [R"U5R X/05$)Nr)_tensor_version)tensor_version_opr&rrr)rnr{r&s rUmethod_attr__version#TensorVariable.method_attr__versions*755oFTT   rWcSSKJn SSKJn U[;a [ S5$U"[ 5RX[ U5/05n[XS5(+nUR(a<[[URU5R[R55 [ U5$![a SnNff=f)Nr/rBuiltinVariableTF)rbuiltinr,all_tensor_attrsr4rRrrPAttributeErrorrr"r#rr!r)rnr{rrr,varret_vals rUcall_obj_hasattrTensorVariable.call_obj_hasattrs &, # ##D) ) !'*88+D12BC %S::G ;; 4;;-889M9MN  (( G s7B22 CCc^^^TRT5(aSTTR5;a[SSTST3STS3STS3/S9 O#TTR5;a[ S TS 35eTS :Xa[ TR 55$[TS T3S5nUbU"T5OSnUbTR(a}TRR5(a^TS ;aUR5(dC[TR[R55 [TRT5UlTRb[!["R$R&T5(a[["R$R&T5n[!US5(a[!XUR)5S5(at["R*R,[XUR)5S5R.;a2[0R2R5[TRT5SS9$Uc TS:waS[6S-4UUU4SjjnU"5nUcTR9TT5nUc[:eU$)NzStrict mode banned opr zGetattr invocation 'z"' in strict mode is not supported.zRemove `zj` from the list of banned ops by setting `torch._dynamo.config._autograd_backward_strict_mode_banned_ops`.rzUnknown property z] during speculating backward, dynamo will insert contiguous call ahead and speculate it againry method_attr_)gradrb overloadsrz9Getting an inplace view on a graph input is not supported)rmsgr8rGc*>SSKJn SSKJn [R TS5nUcg[ U5[RLagURTR5T5nTRbU"TU[TRT5S9$U"TUS9$)Nr/rr)r{r[rr{r[) rrrrr/rrtypesGetSetDescriptorTypecreate_getattr_proxyrrr#)rr static_attrr[rrnr{s rUtry_generic_attr_handling=TensorVariable.var_getattr..try_generic_attr_handlingDs21.224> & $E,F,FF'<? ?$,tf 5t< ' 3    --//55&:S:S:U:U $//,*A*AB C&t{{D9FM ;; "wuyy~~t'D'D.BK((B q 122II**gb,,.:K.L.Q.QQ!~~==%dkk48S> >dfn =/E = =2/0F >))"d3F >% % rWc  UR(d[SSU3S/S9 UR(deURRURRS.nSn[ URR U5nUc[S SU3S /S9 [URR[R55 [U5n[R"U5$![an[SSU3S/US9 SnANSnAff=f) Nz$Unsupported call_id() without sourcezcall_id z6call_id() not supported for sourceless TensorVariable.rrz#Error getting associated real valuezJDynamo encountered an error while trying to get the associated real value.rrrrfrom_excz'call_id() without associated real valuez>Dynamo could not find an associated real value for the tensor.)rrrrrrrrr"rr!ID_MATCHidr4r)rnr{rrrid_values rUcall_idTensorVariable.call_idfs{{ >"4&)T  {{{ii++"))2H2HI'+$ +/ 0@0@%+H ( ( / A"4&)\   dkk,,\-B-BCD23&&x00)  ="4&)1   s$ C)) D 3DD c URS:$)Nr)r_rs rUhas_unpack_var_sequence&TensorVariable.has_unpack_var_sequencesyy1}rWidxesc VSSKJn SSKJn UR 5(a[ UR 5nO@URUS/05n[U[5(de[ UR5nUS:wdS5eUR 5(aUR SnOURUS[R"S5/05n[U[5(dUR5(de[U[5(aURUR 5nOUR#5nUc [%U5nO&[ U5U:XdSUS[ U5S 35e[X5(aeUV s/sHn U"[&XR)5U S 9PM! n n U V s/sH+n UR*"XUR,UR.5PM- sn $UV s/sH$n U"[1U5XR)5U S 9PM& sn $s sn fs sn fs sn f) Nr/wrap_fx_proxy_clsr>r`rzCan't unpack scalar tensors.zCan't unpack a tensor of z rows into a tuple of z elements.r|r{r[)rratorch_functionr?rlenr`rrPr6rr4rSymNodeVariablerF evaluate_exprras_python_constantrangerYrfrom_tensor_varrgrr) rnr{r^rar?size_lensize_varlength dyn_lengthi base_varsrs rUunpack_var_sequence"TensorVariable.unpack_var_sequences /@ ??  499~H''FB;Hh 55 558>>*H1}<<<} ??  YYq\F))"f7G7N7Nq7Q6RTVWJ:770022 3*o66#11"))<#668 =&MEu:' +F83I#e*U_` ' d 9 9  A"-"MMOA"S U/[AU/UQU5Q76nURC5nU"UU5nUREUU[GS!U555 U$s snn f![6a'n[SSUSUSUSU3SUS3/US9 SnANSnAff=f![8a Nf=f)"Nr/)SourcelessBuilderVariableBuilder)can_dispatch_torch_functiondispatch_torch_functionz(Illegal method invocation in strict mode call_method r6z/Dynamo currently does not support this method (z) invocation in strict mode.rry__eq__rFrandom_zTensor.random_ opzTensor.(args= , kwargs=)z This is currently not supported.z'Use the out-of-place version of this opuniform_fromz-Tensor.uniform_ op called with `from` keywordzAvoid using the `from` keyword.method_zUnhandled args for methodz6Dynamo encountered an error while calling the method ``.rTrrc3@# UHoR5v M g7fr)r)rargs rUr-TensorVariable.call_method..Ts#DtMMOOts)$rrrrcrrrCrDrr/rrlistrr#rrRrLrQrPr8rr4r SUPPORTABLErrealize TypeErrorr0rr create_proxyr+rrany)rnr{rrrrrrrrr@is_base_tensor_methodfunc_varhandler_methodrrrealized_kwargsrOerr[rs rUrTensorVariable.call_methods% @X   r " "t/K/K/M'M B&tfAdV1TF!F8D657  '**46 +4 7 (E4&4:2E,F O O%{{* :dkk;#GN-33B d8ST*eTFT$Z$78&   8  474L M M--e4 4 9  +!$wz&!<>=&22  Z Ff$4 G!$wz&!<>5&22   $TWTF+;Dlln"Mnda1iik>n"M'ETE_E!M + &&   }t}f 5 //1r5) && #Dt#D D  I#N 7*4&$qaxH!##'&!,     s<KJ #J?J J J>J99J> K Kc.UR"S/UQ70UD6$r_method_size_stridernr{rrrs rU method_sizeTensorVariable.method_sizeYs''@@@@rWc.UR"S/UQ70UD6$)Nrarrs rU method_strideTensorVariable.method_stride^s''B4B6BBrWrc[U5nS[[S[S[4SjnUS:XaUO[R nUS:wa [ X5nO*US:Xa"UR5(a URnOSnUb#UcU"U5$[R "XR5$URRRRS5=nb}Uc:[ Xa5"5n[U5(dU"[SU555$g[ Xa5"U5n[U5(d[R "[U55$g)NroptionsrGc p[UVs/sHn[R"U40UD6PM sn40UD6$s snfr)r6r4r)rrys rUmake_const_size_variableDTensorVariable._method_size_stride..make_const_size_variablehs=@AB1!((6g6BFM Bs!3r`rc38# UHn[U5v M g7frrrrs rUr5TensorVariable._method_size_stride..s,DVSVVV)r(rrr6r4rrRrr`r[rkrrrrr)rnrrr RetVariablerfakefake_rs rUr"TensorVariable._method_size_stridecs:3   # ,  )- $#A V^ 1 1 AA ={"1~%'..qv66JJOO((,,_= =D J{ ,.'//'u,DV,D'DEE0!,S1'//+223v;??rWcbUR5(a)[R"[UR55$UR R RRS5=nb?UR5n[U5(d[R"[U55$gr) rr4rr*r`r[rkrrnumelrr)rnr{rrs rU method_numelTensorVariable.method_numels} ??  #**7499+=> >JJOO((,,_= =D JZZ\F#F++'..s6{;;rWc^URb [R"UR5$gr)r_r4rrs rU method_dimTensorVariable.method_dims$ 99 #**4995 5rWcrURb*[R"URR5$gr)r\r4ris_floating_pointrs rUmethod_is_floating_point'TensorVariable.method_is_floating_points, :: !#**4::+G+GH HrWc2[R(a,[SSS/[RQ[R QS9 UR RRRS5=nb$[R"UR55$g)Nz0Encountered tensor.is_inference() during tracingr-z&tensor.is_inference() is not supportedrr) r"fake_tensor_disable_inference_moderr FUNDAMENTALINFERENCE_MODEr[rkrrr4r is_inference)rnr{rs rUmethod_is_inference"TensorVariable.method_is_inferences  4 4 JD&22&55  JJOO((,,_= =D J#**4+<+<+>? ?rWcrURb*[R"URR5$gr)r\r4r is_complexrs rUmethod_is_complex TensorVariable.method_is_complexs* :: !#**4::+@+@A ArWrcRUbUR5O[RnURb"[R "X0R;5$UR RRRS5=nb#[R "URUS95$g)Nrr) rgrLcontiguous_formatrdr4rr[rkrr)rnr{rmemory_format_constrs rUmethod_is_contiguous#TensorVariable.method_is_contiguouss (  , , .((     )#**+>BTBT+TU Ujjoo**..??d L#**""1D"E rWF non_blockingc ^UcTRb[TR[R5(a[ U4Sj[ R "555nTRRS:Xa#[R"SUR35$[R"STRRSUR35$Ub[[UR555S:XasUR5n[R"[U55nSSK Jn U(aSU0UEnU"UURR "S S /[#TU/U5Q765$g) Nc3P># UHupTRU;dMUv M g7fr)r\)rrrrns rUr-TensorVariable.method_type..s#9daTZZ1_9s& &cpuztorch..ztorch.tensortyper/rrrr)r\rPr]rLnextr.rrr4rrSr$rgrrrrr+) rnr{r\rrr tensortype tensor_typetensor_type_constrs ` rU method_typeTensorVariable.method_typesT M &4;; 551779J{{5('.. 8K8K7L/MNN'..T[[--.a 0C0C/DE  D113459KK  224K 0 7 7K8H I  .(,A&A  &&!'.?'@&I rWclsr?cb[U[5(awUR(afSSKJn UR 5nUR "XXBR5nURRR[5U[S9 U$[SUSUS3SS/[RQS 9 g) Nr/r>mutation_type_clszHArgument of `as_subclass` must be a non-dispatcher-style tensor subclassz .as_subclass(rCurrently not supportedz:Avoid this call or move it outside `torch.compile` regioner)rPTensorSubclassVariablerrcr?rgrir side_effects _track_objobjectr0rrr)rnr{rr?py_clsr1s rUmethod_as_subclass!TensorVariable.method_as_subclasss c1 2 2szz D++-F.>>&**C II " " - -#1E . J^fM#a01L"..  rWc[UR[R5(aHURRS:waURROSn[ R "U5$g)Nr)rPr]rLrindexr4r)rnr{rs rUmethod_get_device TensorVariable.method_get_devicesM dkk5<< 0 0)-)9)9U)BDKK%%E#**51 1rWcV[R"URR5$r)r4rr\itemsizers rUmethod_element_size"TensorVariable.method_element_sizes&&tzz':':;;rW)forcerNumpyNdarrayVariablec,[R(d[SSUS3SS/S9 [(d[SSUS3SS /S9 UR[ R :wa[S URS 35eU(aWUR5(aBURUS /05nURRS SUR5405nO)URR"S S/[X/05Q76n[RX5$)Nz%Tensor.numpy() with trace_numpy=Falserz numpyzW`Tensor.numpy()` was called, but the `trace_numpy` configuration was manually disabled.zUSet `torch._dynamo.config.trace_numpy = True` to allow Dynamo to trace through NumPy.rz&Tensor.numpy() without NumPy installedz_`Tensor.numpy()` was called, but the NumPy library is not available in the current environment.z5Ensure NumPy is installed in your Python environment.zcan't convert z4 layout tensor to numpy. Use Tensor.to_dense() firstdetachrrview_as)r trace_numpyrnpr^rLstridedrrgrrrrr+rr)rnr{rtr[s rU method_numpyTensorVariable.method_numpy!s!! ?&tfF375 r @&tfF3?L   ;;%-- '  -ab  U--//  Xr26AII**=%!**,RTUEII**y+S[S[RRS[4UU 4SjjnUR [R [R[R[R4;a[SSTS3SS /S 9 UR5S :XaU"X5$UR5S :Xa)[U5VVs/sHup4U"XAU5PM snn$[U5VVs/sHup5T"XQUS 9PM snn$s snnfs snnf)NrnrrGc2>T"TUR55$r)item)rnrr{rs rUwrap:TensorVariable.method_tolist..tolist..wrapOs$NN$rWz'Tensor.tolist() with non-integer tensorrz to_listzLDynamo currently does not support tracing `tolist()` on non-integer tensors.z[Ensure the input tensor to `tolist()` is an integer type (e.g., int8, int16, int32, int64).rrr/)r) rrLfxProxyr2r\int8int16int32int64rr enumerate) rrr rnval sub_tensorrntolistr{rs rUr,TensorVariable.method_tolist..tolistNs   ?   ||    $ E*4&9!9B zz|q F..zz|q >G>OP>OFASA,/>OPP&/v%6%6MAzq\:%6 Qs D5Dr) rrrLrQr r rrrrkrr2r)rnr{routrrs`` @@rU method_tolistTensorVariable.method_tolistKs~*! 5<<! EHHNN! sTRUY! ! F%%**?;VT]]_-$$R--rW vars_itererror_on_non_leafcSSKJn /n[5nUHn[U[5(dMUR (dM-UR (aU(a[SSU3SS/S9 MWMYUR(a[URU5(aU(a[SS U3S S /S9 MMURRnXu;dMURU5 URU5 M U$) a Collect unique leaf tensors from vars_iter for backward. Only collects leaf tensors (no grad_fn). Non-leaf tensors are skipped (or error if error_on_non_leaf=True) because when auto-detecting inputs, we must not stop gradients at non-leafs - they are intermediates, and the real leaf tensors (parameters) are further up the autograd graph. Deduplicates by proxy.node. Returns list of unique leaf tensor variables. r)SyntheticLocalSourcezbackward() with non-leaf tensorz-backward(inputs=[...]) with non-leaf tensor: zBbackward(inputs=[...]) with non-leaf tensors is not yet supported.zIOnly pass leaf tensors (parameters, graph inputs) to backward(inputs=...)rz'backward() with in-graph created tensorz5backward(inputs=[...]) with in-graph created tensor: z^backward(inputs=[...]) with tensors created inside the compiled function is not yet supported.zSOnly pass tensors that are inputs to the compiled function or captured from outside) rrsetrPrYrbrqrr[rkaddappend)rnrrrrO seen_nodesr1rks rU_collect_backward_inputs'TensorVariable._collect_backward_inputsus 2), C#~..33D3D3D??(%$E&STWSX$Y(l k# )z#**>R'S'S(%$M&[\_[`$a)F!v# )99>>D-"t, c*EH rWgradient retain_graph create_graphinputsc :[R(d[SSUSUSUSUSU3 SS/S9 UR(dUR(d [ S5eUS LnU(aa[ URRURRR55nURU5nU(d[$OR[U[R5(a UR OU/n URU S S 9nU(d [S S SS/S9 S["R$"X50n UbX:S'UbXJS'["R$"X5n X /n UbU R'U5 ["R$"U[(R*R,5n U R/XU 5nSSKJn UR5USS S9nUR7U5 ["R$"U[(R8R:R<R>5nUce[AU5HEunnURCUS["R$"UU5/05nUR/UUU/05 MG UREU5 ["R$"US 5$)aD Trace tensor.backward() by rewriting as autograd.grad() + accumulate_grad. Implementation: 1. Collect leaf tensors to compute gradients for 2. Call autograd.grad(loss, inputs) to compute gradients 3. For each leaf tensor, call accumulate_grad to update .grad Non-leaf tensor handling: - Auto-detect (inputs=None): Non-leaf tensors are silently skipped. This matches eager where only leaves get .grad. - User-provided (inputs=[...]): Errors if any non-leaf tensor is found. While eager backward(inputs=[non_leaf]) works, Dynamo cannot trace it because the accumulate_grad polyfill accesses .grad, and Dynamo creates a generic GetAttrVariable for .grad on non-leaf tensors (instead of a TensorVariable), which cannot be used in tensor operations. TODO: Support non-leaf tensors by fixing .grad access on non-leaf in Dynamo. z"Unsupported Tensor.backward() callrz backward r6z]Dynamo currently does not support tracing `Tensor.backward()` when trace_autograd_ops is off.z)Set torch._dynamo.trace_autograd_ops=Truerz8tensor does not require grad and does not have a grad_fnNT)rzbackward() with empty inputsz8backward(inputs=[...]) resulted in no valid leaf tensorsz?backward(inputs=[...]) requires at least one valid leaf tensor.zOEnsure at least one tensor in inputs is a leaf (requires_grad=True, no grad_fn) allow_unusedr$r%r/)GradModeVariableF) initialized __getitem__)#rtrace_autograd_opsrrbrqrrrleaf_var_creation_orderinput_source_to_varvaluesr!r3rPrBaseListVariablerr2rrrLautogradr8r ctx_managerr)renterrHinductoraccumulate_grad_defaultrrexit)rnr{r#r$r%r& auto_detectall_vars input_vars provided_vars grad_kwargs inputs_var grad_argsautograd_grad_fn grads_varr) grad_mode_varaccumulate_grad_fnidx input_vargrad_is rUmethod_backwardTensorVariable.method_backwards6(( <&tfJxj,qQ]P^^_`f_gh{BC  !!$*:*:#J n  11 --446H66x@J.- fi&@&@AA X  667J:V ai &'<'"4&(8axH^  rWc .[SSUSUSU3S/S9 g)Nz2Unsupported Tensor.sparse_resize_and_clear_() callrz sparse_resize_and_clear_ r6zNDynamo currently does not support tracing `Tensor.sparse_resize_and_clear_()`.rrrs rUmethod_sparse_resize_and_clear_.TensorVariable.method_sparse_resize_and_clear_s, H"4&(B4&&Rh  rWc l[U5S:a%[SSUSUSU3S/[RQS9 g)Nr/zUnsupported Tensor.set_() callrz set_ r6zhDynamo currently does not support tracing `Tensor.set_()` overloads that include more than one argument.r)rdrrrrs rU method_set_TensorVariable.method_set_sL t9q= 8&tfF4&&BA6)556  rW)alphaotherrcUb`[R(aK[R"[R 5R XU/05nURUSU/05$gNadd_)rrrrrrLmulrr)rnr{rrrOs rU method_add_TensorVariable.method_add_sX  !D!D;;EIIFTTENBF##B"= =rWc6Ub[R(a[R"[R 5R XU/05n[R"[R5R XU/05nURUSU/05$gr) rrrrrrLdivrrr)rnr{rnrorFrOs rUmethod_addcdiv_TensorVariable.method_addcdiv_s  !D!D;;EIIFTTg&F;;EIIFTTUORF##B"= =rWrc[R"[R5R XU/05n[R"[R 5R X/05nUR US/05$)Nr)rrrLeqrrr)rnr{rrOs rUmethod___contains__"TensorVariable.method___contains__sm77AOO s R 77 BPP " !!"fb"55rWc b^ ^ UVs/sHoDR5PM snm UR5VVs0sHupVXVR5_M snnm S[S[4U U 4SjjnSUlSSKJn U"UUR R"SU/[U/05Q76S9$s snfs snnf) NrrGc(>UR"T0TD6$r) redistributer args_as_valuekwargs_as_values rUredistribute_fn_with_prim_typesKTensorVariable.method_redistribute..redistribute_fn_with_prim_types2s>>=DOD DrWprim_redistributer/rrr<) rgrrrSrrrrr+) rnr{rrrrrrrrrrs @@rUmethod_redistribute"TensorVariable.method_redistribute's:>>A--/> AGP12244P Es Es E E4G'0*))((/#D62.  ?Ps B&B+c P^ ^ S[S[4SjnURS[5n[ U[ R 5(a+[ R"[5RX/05nURS5bXSS'UVs/sH od"U5PM snm UR5VVs0sH upxXt"U5_M snnm S[S[4U U 4Sjjn SU l SS K J n U "UURR"S U /[!U/05Q76S 9$s snfs snnf) NvtrGcx[U[R5(a UR$UR 5$r)rPrUserDefinedObjectVariablerFrg)rs rUextract_python_value.extract_python_valueNs-"iAABBxx((* *rWgrad_placementsrc(>UR"T0TD6$r)to_localrs rUto_local_fn_with_prim_typesCTensorVariable.method_to_local..to_local_fn_with_prim_typesas::}@@ @rW prim_to_localr/rrr<)r2rrr3rPrrr,rrrrSrrrrr+) rnr{rrrrgrad_placements_vtrrrrrrrs @@rUmethod_to_localTensorVariable.method_to_localCs- +_ + + $ZZ(9;QR ()*M*M N N!*!:!:5!A!O!O(""  ::' ( 4(:$ %:>?$Q-a0$? BH,,.Q.$!12155.Q A3 A3 A A0?#,*))((+#D62.  @Qs D5D"c0UR"US/UQ70UD6$)N register_hook_method_register_hookrs rUmethod_register_hook#TensorVariable.method_register_hookrs! ))"oOOOOrWc0UR"US/UQ70UD6$)N"register_post_accumulate_grad_hookrrs rU)method_register_post_accumulate_grad_hook8TensorVariable.method_register_post_accumulate_grad_hookzs- )) 4 7; ?E  rWhookc D^^^^^UR(Gd1[R(aURR U5umnS[ R S[RSS4UU4SjjnSSKJ n UR5nSURRS'U"UURRS UXt4055$UR5Rn[URR!55n ["R$"[&5n U R)XU/05n U R5Rn /m[+5mS [ R,R.S [ R,R.SS4UUU4S jjmT"X5 Un THnU R1U5 Un M U HnUR3X5 M [5U [65(deU R5UlUR;U5 ["R<"["R>RA5S 9$["R<"["R>RA5S 9nURRBREXUT5 U$)Nrbw_staterGc h>[UT5nU"[R"[[UTS95 g)N)rPr hook_name)rR functoolspartialrr )rrrrrs rU_register_hook_trampolineGTensorVariable._method_register_hook.._register_hook_trampolines8%,FD$9M!!)))<%-&/  rWr/rThas_backward_hookrrkstop_atc>UT;dXLagURH6n[U[RR5(dM.T"X!5 M8 UR R 5H6n[U[RR5(dM.T"X15 M8 TRU5 TRU5 gr) rrPrLr Noderrr/rr)rkrrkwarg collect_deps nodes_to_movevisiteds rUr:TensorVariable._method_register_hook..collect_depss7?do99C!#uxx}}55$S2%"[[//1E!%77$U42 D!$$T*rWrj)#rr compiled_autograd_enabledradd_backward_state_hookrLrQ BackwardStaterrrrkrrruserskeysrAutogradFunctionVariabler call_applyrr rrreplace_input_withrPrYr[rRemovableHandleVariablebaser1rr)rnr{rrbw_state_proxyrr self_proxy tensor_nodeusers_to_replaceapply_hook_varrOtensor_prime_node insert_pointrkuserhandle_variablerrrrs ` @@@@rUr$TensorVariable._method_register_hooksD {{{!::,.II,M,Md,S) > !LL 4E4S4S   "3!]]_ <@ $$%89$II**'1#4 P--/..K#K$5$5$:$:$<= '??@RSN#..r$<DF!' 1 6 6  24M*-%G +588== +588== +T + + * 8'L%##D)# & )'' G)fn55 55*DJ  ' ' + 44'nn==? $;;#..99;  ,,T$OrWcUSLaUR5nUR5RRSRU:wa[ SSUS3S/S9 gU$)NTrz(Unsupported Tensor.requires_grad_() callrz requires_grad_zaDynamo does not support changes to a Tensor's `requires_grad` through calling `requires_grad_()`.r)rgrrkrrbr)rnr{rbs rUmethod_requires_grad_$TensorVariable.method_requires_grad_sf  $)<<>M ==?   $ $_ 5 C C} T B&tfO<F  KrWcF[SSUS3SS/[RQS9 g)Nz'Unsupported Tensor.share_memory_() callrz share_memory_zTDynamo does not support Tensor.share_memory_() which modifies tensor storage for IPCz7Move share_memory_() call outside the compiled region. r)rrrrms rUmethod_share_memory_#TensorVariable.method_share_memory_s2="4&7nI"..  rWc[U5S:Xa[US[5(d&[U5S:a*[SU55(aUR USX#5$g)Nr/rc3# UH:nUR5=(a [UR5[5v M< g7fr)rFrPrgr)ras rUr,TensorVariable.method_new..4s8A$$&R:a6J6J6Lc+RRsAA new_empty)rdrPr6allrrs rU method_newTensorVariable.method_new*s^ INz$q'<@@ IN ##B TB BrWc~[XR5RRSR 55$r)UntypedStorageVariablerrkruntyped_storagers rUmethod_untyped_storage%TensorVariable.method_untyped_storage<s4& --/&&++O<LLN  rWc~UR(d,URRRU5 SUlgr)rir[rk_rename)rnrs rU set_name_hintTensorVariable.set_name_hintCs,  JJOO # #D ) $D rWcTUR5RRSSL$r)rrkrrms rUis_python_hashable!TensorVariable.is_python_hashableIs&}}##((9EErWcb[UR5RRS5$r)hashrrkrrms rUget_python_hashTensorVariable.get_python_hashOs$DMMO((--o>??rWc[U[5(dgUR5RRSnUR5RRSnX#L$)NFr)rPr2rrkr)rnrrbs rUis_python_equalTensorVariable.is_python_equalRsR%11 MMO % %o 6 NN  ! ! & & 7v rW)rirprgr]r\rqrdrercrfr^r_r[rbrar)NF)F)NNNN)rGNT)r{r;rGr)wrS __module__ __qualname____firstlineno____doc__r2_nonvar_fieldsrLrQr'r r r\r]r^rrJrrrrwrrrstrrrrr staticmethoddictrrrrrrr4rrrrr rr rr r#r(r3rQrYr\rrrprwrpropertyr`rDrErrrrrmethod_nelementrmethod_ndimensionrrrrrrrrrrrr r!rFrJrPrVrcacherOrgr5rlrurrrrrrrrrrrrrrrrrrrrr rr __static_attributes__ __classcell__rys@rUrYrYs G   ' '!N&B B.)-)-%)$(#'/xx~~'/{{ '/  '/  '/'/'/'/'/'/'/'/S#X%'/c3h$&'/ d{!'/"Tk#'/$%'/& ''/'/TFJ  )  7;d{    E3:E , #,d , ,  ,(;C;%((..T4/%,,/4S>//bHB)HB14HB HBT7#:77 $;RV@V %<SWAW %<SWAW ) D 8$;88) D ) D ) D ) D  *A h (#:(( 4) 4 D  4 '> ? )))14) )R !  #  Tz  t$& #&& &  &B  #    "      #    "      #    "      #    "     #"   6)-  #   %  4  ()-  #""  %  4 $ 6) 60? 6  6 #  "    8- #- - " -  - ^P #PP" P  P #  "    H)H14H:P "  h   #"  4  $ ) ! #$ FDF @@VrWrYc h^\rSrSrSrSS1\R krS\4Sjr\ S SS S\ S\ S-S \ SS 4 S jj5r S\ S\ S \ SS4U4Sjjr S\ 4SjrS\4SjrS\ 4SjrSS S\ S\4SjrS S\SS\\-\-4SjjrSSS\S\\S \\\4S\4 SjrS\4SjrS\4SjrS\S\4SjrSrU=r $)!reiZz Represents a symbolic scalar, either int, float or bool. This is most commonly used to handle symbolic size computation, e.g., tensor.size(0), but it is also used to handle logic like float_tensor.item() or unspecialized float inputs. r[sym_numrGc,[UR5$r)rr rms rUrSymNodeVariable.debug_reprgsDLL!!rWNr{r<rr2c DUc[URU5nSURR;aURRSU:Xde[URU5 [ U[ R [[45(aB[ U[ R 5(a [U5OUn[R"U5$[X#40UD6nURRS:wa%URRRU5 U$)Nrrt)r&rkrr-rPsympyIntegerrrJr4rrerxrcurrent_tracerrecord_tensor_or_symint_vt)rr{r[r rrs rUrSymNodeVariable.createjs ?$UZZ4G ejjoo -::???3w> >>%**g. g sD9 : :&0%--&H&Hc'lgG#**73 3e88 ::==M ) II $ $ ? ? D rWrrc L>[TU]"S0UD6 XlX lSUlgNru)rvrwr[r  _tensor_var)rnr[r rrrys rUrwSymNodeVariable.__init__s% "6"  26rWc[UR[5(a URRR$[ UR5$r)rPr rrkpytyperrms rUrSymNodeVariable.python_types8 dllH - -<<$$++ + % %rWcgrrurms rUis_symnode_likeSymNodeVariable.is_symnode_likerrWcUR$rrrms rUrSymNodeVariable.as_proxyrrWr\c URcQ[R"U[R5R X/S[R"X505UlUR$)Nr\)r+r2rrL scalar_tensorr)rnr{r\s rU as_tensorSymNodeVariable.as_tensors[    #.44E'' mB/2G2G2R(ST  rW output_graphr:c[UR5$![a^n[RR R R(ae[[RS[U53SS9eSnAff=f)Nz5Consider annotating your code using torch._check*(). constrain_as_size_example) case_name) rr rrLr  experimental_configno_data_dependent_graph_breakrr ANTI_PATTERNr)rnr9rs rUrfSymNodeVariable.evaluate_exprsi  - -* xx$$,,JJ**GAxP5   s A?AA::A?r;rrc rSSKJn U"UURR"SU/[ U/UQU5Q765$)Nr/rr)rrrrr+)rnr{rrrrrs rUrSymNodeVariable.call_methodsF +  II " " #D=4=&9   rWcgrrurms rUr"SymNodeVariable.is_python_hashablerrWc4[UR55$r)rrfrms rUr SymNodeVariable.get_python_hashsD&&())rWrc[U[5(a!UR5UR5:H$[U[5=(a! UR5UR 5:H$r)rPrerfr2rg)rnrs rUr SymNodeVariable.is_python_equals] e_ - -%%'5+>+>+@@ @ uo . C""$(@(@(BB rW)r+r[r r)!rSrrrrr2rrr classmethodrrrwrrrJr1rrYr7r rfloatrfrrrrr rr rrrs@rUrereZs|   ' 'N "C" #  't     ,7c7C73747&T& # 7   7; $]3  e   #  '  S/)*    $D**  V   rWrec ^\rSrSrSr\SSS\RRS\ SS4Sj5r SSS \ S\ 4S jr \S \ S \\ S \\ \ 4S\\\ \\ \ 444S j5rSSS \ S \\ S \\ \ 4S\ 4 U4SjjrS\4SjrSrU=r$)rizq Represents a np.ndarray, but backed by torch Tensor via torch._numpy.ndarray. Use this for Tensor.numpy() call. r{r;r[rrGc .SSKJn U"S[UUS.UD6$)Nr/r`rbru)rrar)r{r[rras rUrNumpyNdarrayVariable.creates, /  +    rWrc^^^^^ ^ SSKJm SSKJm SnTR 5R R Sn[R"U5nS[4UU UUU 4SjjnTS;aETRRS T TR 5T405n[RTU5nOTS ;a [R"[UT55$TS ;aI[![UT5=n5(d&[R"[#S U555$U"5$TS :XaB[!UR$=n5(d[R"['U55$U"5$TS;a[)SSTST3STS3/S9 OTS:Xa[)SSTST3STS3/S9 Uc[*eU$)Nr)numpy_attr_wrapperr/rrrGc n>T"TTRRSTTR5T4055$)Nr)rrr)rrPrnr{rsrUinsert_into_graph;NumpyNdarrayVariable.var_getattr..insert_into_graphs;  &&#%7$--/49PRT rW)Trealimagr)r_r)shaperac38# UHn[U5v M g7frrrs rUr3NumpyNdarrayVariable.var_getattr..s4GQSVVQrr`)rflagsr\z$Unsupported ndarray attribute accessrr6z3Dynamo currently does not support tracing `ndarray.rr __version__z&Unsupported ndarray.__version__ access)rNrPrrrrkrtnpndarrayr2rrrrr4rRrrr`rrr) rnr{rrOrexample_ndarrayrRr[rrPrs ``` @@rUrQ NumpyNdarrayVariable.var_getattrs /* ,,11/B ++m4 ?   ( (II**"$' E *00U;F) )#**7?D+IJ J ( (#$)G$GAHH'..u4GQ4G/GHH$& & V^#)=)=$=A>>'..s1v66$& & / / >&tfAdV4QRVQWWYZ   ] " @&tfAdV4QRVQWWYZ   >% % rWrrrcUS:Xa9SSS.nUR5VVs0sHupEURXD5U_M nnnX4$s snnf)Nclipminmax)a_mina_max)rr)rrrr kwargs_renamerrs rU patch_argsNumpyNdarrayVariable.patch_args(sO 6>&+eSSKJn SSKJn UR X#U5up4US:XaSSKJn SnSU;aUSnO[U5S:aUSnUSL=(a URS 5n [X5=(a UR[Ln U (dU (a%U"S S US US US U3S /[RQS9 US;a[T U]=XX45$US;aU"SS US US US U3SUS3/S9 UR R""SU"U5/[%U/['U5-U5Q76n [(R+X5$)Nrr)numpy_method_wrapperastyper/r+r\rOzndarray.astype(object)rr6z`ndarray.astype('O')` or `ndarray.astype(object)` is not supported by torch.compile, as there is no equivalent to object type in torch.Tensor. This will be executed eagerly.r)__len__r`r__iter__)tostringtobytes __delattr__zUnsupported ndarray method callz `ndarray.z&()` is not modelled in `torch._numpy`.r)rrrNrjrgr.r,rdis_constant_matchrPrPrrrrvrrrr+rrr) rnr{rrrrrrjr, dtype_arg is_object_stris_object_typer[rys rUr NumpyNdarrayVariable.call_method1sy (0t6:  8  0I& "7O TQ G %T1Vi6Q6QRU6VM96Q9<<6;Q 4*4&$qaxH9;-99:  < <7&r> > 9 9 9&tfAdV1TF!F8D'v-ST    &&   & vT 2F ;  $**255rWc<[b[R$[$r)rr]rrms rUr NumpyNdarrayVariable.python_typees >:: OrWru)rSrrrrrrLr r rrrr2rQrrrrgrrrrrrs@rUrrs   #  ,1HHNN  GJ      E5ESE_EN!/2<@oAU[TU]"U40UD6 X lX0lgr)rvrwr{r|)rnr[r{r|rrrys rUrw$UnspecializedPythonVariable.__init__ws" )&)"&rWtensor_variablecH[S0[UR5DUUS.D6$)Nr}ru)rzr__dict__)rrr{r|s rUfrom_tensor_variable0UnspecializedPythonVariable.from_tensor_variables/+ ?++, #  rW)r|r{r)rSrrrrrYrrLr r rKrrJrrwrJrrrrs@rUrzrzls   & &N)- 'xx~~ '3;% '  '  '  ' ' !  '  3;%    '    rWrzc^\rSrSrSrS1\R krS\RRS\ SS4U4Sjjr \ S \SS4S j5r S rU=r$) FakeItemVariableizAn unspecialized python variable which prevents access to the underlying raw value. This is needed if item is called on a FakeTensor.r|r[rrrGNc X>URSS5n[TU]"U40UD6 X0lg)Nr|F)poprvrwr|)rnr[rrr|rys rUrwFakeItemVariable.__init__s+jj6  )&)&rWrc>[S0[UR5D6$r*)rrr)rrs rUr%FakeItemVariable.from_tensor_variables A$'?'?"@AArW)r|)rSrrrrrYrrLr r rrwrJrrrrs@rUrrsl9   & &N 'ehhnn''' B,B BBrWrcL\rSrSrSSS\\S\\\4S\4SjrS\ 4Sjr S r g ) rir{r;rrrrGc ,SSKJn URRnU[R RLaSn[ U5S:XaUUSR5(a=[ U5S:Xa.USnUR"XURUR5nOd[SURSUSUS3SS /[RQS 9 O3[R"X5RX/[!U5-U5nUceURR"nU[R R"La'[R"X5RX/U5 UR$R&R)[+5U[,S 9 U$) Nr/r>rzCCalling subclass default constructor with more than tensor argumentrrrrzFAvoid this constructor call or move it outside `torch.compile` regionerr)rcr?rF__new__rLrQrdrrirrrrr2rrrrwrrrrr0) rnr{rrrr?new_funcr1data init_funcs rUr$TensorSubclassVariable.call_functionsm A::%% u||++ +C4yA~$q'"3"3"5"5#f+:JAw3BBdjj$++a#zzl&ixqI 92+66 "''5CCFT$Z'CJJ''  ELL11 1  ! !" 0 > >r5& Q )) Hc-A *  rWcUR$rrTrms rUrg)TensorSubclassVariable.as_python_constantrrWruN) rSrrrrr2rrrrrgrrurWrUrrsI/ #/'/S/)* /  /bDrWrc ^\rSrSrS1\R krS\S\RS\ SS4U4Sjjr SS S \ S \ \S\ \ \4S\4 U4S jjrSS jrSrU=r$)rir from_tensorrrrGNc >>[TU]"S0UD6 XlX lgr*)rvrwrr)rnrrrrrys rUrwUntypedStorageVariable.__init__s! "6"&*rWr{r;rrc >US:XaU(dU(a&[UUS[U5S[U5S35 URR5n[ U5(d[ R "[U55$SSKJ n SSK J n U"UURRS UURR54055$US :Xa[U5S:XaU(a[XS [U5S35 URRS [ R"R$R&URR5US R5405 U$[(TU]UXX45$) Nr`z0 args and 0 kwargsz args and z kwargsr)untyped_storage_sizer/rrresize_z0 kwargsr)r,rdrr`rr4rrexternal_utilsrrrrrrrrLrHr4resize_storage_bytes_rvr) rnr{rrrrrOrrrys rUr"UntypedStorageVariable.call_methodsL 6>v#)4ykCK=@  '',,.F#F++'..s6{;;A2$II**',))2246  9 Ta#BjS[M:QR II " " ""88!!**,d1g.>.>.@A   Kw"2T::rWclU"UR5 URS5 URS5 g)Nrrr load_methodrrncodegens rU reconstruct"UntypedStorageVariable.reconstruct s-  !-.ArW)rrrr9rGN)rSrrrr2rrYrLUntypedStoragerrwrrrrrrrrs@rUrrs  ' 'N +# +++ + +  ++; #+;+;?# +; S/)* +;  +;ZrWrcB^\rSrSrS\S\SS4U4SjjrS SjrSrU=r $) rHi! rrrrGNc 2>[TU]"S0UD6 Xlgr*)rvrwr)rnrrrrys rUrwDataPtrVariable.__init__" s "6"&rWclU"UR5 URS5 URS5 g)Ndata_ptrrrrs rUrDataPtrVariable.reconstruct* s,  !J'ArW)rr) rSrrrrYrrwrrrrs@rUrHrH! s.'#''  'rWrH)rrloggingrTrar]r=collections.abcrr contextlibr itertoolsrrtypingrr r r r$ torch._numpy_numpyr\torch.fxrL torch.random torch._dynamor torch._library.opaque_objectr torch._subclasses.meta_utilsr%torch.fx.experimental.symbolic_shapesrrrrrtorch.utils._python_dispatchrr-rrr_trace_wrapped_higher_order_oprrrrrrrrrr guardsr!r"rr#rNr$r%r&r'r(r)r*r+r,r-r.rr0r1r2constantr3r4listsr5r6 script_objectr7 user_definedr8numpyrModuleNotFoundErrortorch._dynamo.codegenr9torch._dynamo.output_graphr:torch._dynamo.symbolic_convertr;r< functionsr=rcr? getLoggerrSr_gtltgelerneis_is_notsupported_tensor_comparison_opssupported_const_comparison_opssupported_comparison_opsrfromkeysr/%supported_tensor_comparison_op_values$supported_const_comparison_op_valuesrrJrVr TensorBaserrQr/rYrerrzrrrrHrurWrUrsL" ."99 +A6G33:O0    JI>542 /6 0<!   ++ ++ ++ ++ ,,oo # ,,oo ++ ++ " %$)- #**,)%(,}}"))+($ &T88&&//%,,2G2GGE_EP6q oq h[>[|# .# LB~B*353lA_AH o aF Bs.JJJ