ȅitSrSSKJr SSKrSSKrSSKrSSKrSSKrSSKrSSK r SSK r SSK r SSK r SSK r SSKrSSKrSSKrSSKJr SSKJr SSKJrJrJrJr SSKrSSKJr SSKrSSKJr SSK J!r! SS K"J#r# SS KJ$r$ SS K%J&r& SS K'J(r(J)r) S SK*J+r+ S SKJ,r,J-r-J.r. \(aSSK/J0r0J1r1 SSK2J3r3 SSK4J5r5 \ Rl"\75r8\"S5r9\"S5r:\:Rw5r<\<(aSSK=r>/r?Sr@SrA\<(ah/SQr?\>RRRR5RSS5rASR\?Vs/sH nSUS3PM sn5r@/SQrH"SS5rIS?SjrJS rK"S!S"5rL\RS?S#j5rNS$S%.S@S&jjrOS$S%.S@S'jjrPS?S(jrQSAS)jrR"S*S+\S5rTSBS,jrUSCSDS-jjrVSES$S$S..SFS/jjjrWSGS0jrXSHS1jrYSIS2jrZSES$S$S..SJS3jjjr[SKS4jr\SLS5jr]\]"\R5r_\]"\R"S655ra\]"S5rb\]"S$5rc\]"S$5rd"S7S85re"S9S:5rf"S;S<5rgSMSNS=jjrhSOS>jrigs snf)Pa Debug utilities for TorchDynamo compilation and execution. This module provides various debugging tools and utilities for TorchDynamo, including: - Minification support for reducing test cases while preserving bugs - Input/output handling via InputReader and InputWriter for reproducible testing - Accuracy checking between original and compiled models - Neural network module string conversion via NNModuleToString - Profiling tools and system information collection - Buck build system integration for Meta-internal testing Key classes: - InputReader/InputWriter: Handle serialization of model inputs/outputs - NNModuleToString: Converts nn.Modules to string representations - BuckTargetWriter: Manages Buck build system integration ) annotationsN)Counter) import_module)AnyOptional TYPE_CHECKINGTypeVar)Tensor) rand_strided)normalize_path_separator)is_float_dtype)StorageWeakRef)ContentStoreReaderContentStoreWriter)config) clone_inputs get_debug_dir warn_once)CallableSequence)tqdm)UntypedStorageTztorch._inductor.config)z1//caffe2/torch/fb/sparsenn:sparsenn_operators_gpuz-//caffe2/torch/fb/sparsenn:sparsenn_operatorsz///deeplearning/fbgemm/fbgemm_gpu:sparse_ops_cpuz+//deeplearning/fbgemm/fbgemm_gpu:sparse_opszfbcode:// ztorch.ops.load_library("z"))buck2runz@mode/dev-nosanc6\rSrSrSSjrSSjrS S SjjrSrg) BuckTargetWriterYc,[RR[RRU55uUlUlUR R SS5UlURR SS5SUR3UlURURRS5SUlURSSUlURnX"RS5SSSnSUS UR3Ul g) Nz.pyr/.zfbcode.zfbcode/r:) ospathsplitabspathsubdirpy_filereplacetargetfind cmd_line_path)selffilenametmps S/home/james-whalen/.local/lib/python3.13/site-packages/torch/_dynamo/debug_utils.py__init__BuckTargetWriter.__init__Zs$&GGMM"''//(2K$L! T\ll**5"5 {{**345Qt{{mD IIdiinnY79: IIabM kk((9%'(,!#a }5c SR[Vs/sH nSUS3PM sn5n[R"SURSUR S[ SUSURS 3 5$s snf) Nrz "z",za load("@fbcode_macros//build_defs:python_binary.bzl", "python_binary") python_binary( name="z", srcs = ["z"], compile = False, deps = [ "//caffe2:torch", "//caffe2:libtorch", "//caffe2/functorch:functorch", "//triton:triton", "z", ], cpp_deps = [ z ], main_module = "z", par_style = "xar", ) )join extra_depstextwrapdedentr/r- cur_targetr))r2xextra_cpp_depss r5buildBuckTargetWriter.buildhsz#Jz!is"$5z#JK  ;;-ll^  II;#   $KsA3cd[RRURS5n[ US5nUR UR 55 SSS5 [UR/-nU(a%[RSSRU55 U$!,(df  NP=f)NTARGETSwzFFound an example that reproduces the error. Run this cmd to repro - %s ) r(r)r:r,openwriterABUCK_CMD_PREFIXr1logwarning)r2 print_msg target_filefd cmd_splits r5rHBuckTargetWriter.writesggll4;; : +s #r HHTZZ\ "$$t'9'9&::  KKX# $ #s B!! B/)r1r)r-r,r/N)r3strreturnNonerRrQ)T)rLboolrR list[str])__name__ __module__ __qualname____firstlineno__r6rArH__static_attributes__r8r5r!r!Ys 6 4  r8r!c([RR[5S5nUc-[R "5S[ R"53n[RRU5(d[R"USS9 U$)Nminifierz /minifier_T)exist_ok) r(r)r:rtempfile gettempdirgetpassgetuserexistsmakedirs)r)s r5 minifier_dirrfse 77<<  4D |%%'( 7??3D2EF 77>>$   D4( Kr8c\rSrSr\R R \R R\R R\R R\R R\R R\R R\R R\R R\R R\R R \R R"\R R$\R R&\R R(\R R*\R R,\R R.\R R0\R R2\R R4/r\SSj5r\SSj5rSrg)NNModuleToStringc[5nUR5H5up#[U5[R;dM$UR U5 M7 [ U5S:a[RSU5 g)Nrz-We have not tested reprs of some modules - %sT) setnamed_childrentyperi safe_reprsaddlenrJrK)gm cant_convert_modules r5can_convert_to_string&NNModuleToString.can_convert_to_strings`u **,IAF|#3#>#>>  (- | q KKG Vr8c$SSKJn Sn[R"S5nUR 5HYupEUR 5n[ UR5S5nUbUR(aUS3nX2S-SUSUS 3- nM[ URR5HupU cM U R5[::a(SS K Jn U R[:de[!U 5n Oh["R$"U 5(a'S ['U R(5S U R*S 3n O&S['U R(5S U R*S 3n U R(aU S3n UUS-SUSU S3- nM UR,R5HWupU cM SnU R(aSnS['U R(5S U R*US3n X2S-SU SU S 3- nMY X1"UR.S5S 3- nU$)Nr) _addindent z from torch.nn import * class Repro(torch.nn.Module): def __init__(self) -> None: super().__init__() z.cuda()zself.z = r) PRINT_OPTSz torch.randn(z, dtype=)ztorch.randint(1, size=zself.register_buffer('z', z) rz, device="cuda"ztorch.nn.Parameter(torch.randn(z))rg)torch.nn.modules.moduleryr<r=rm__repr__next parametersis_cuda_buffersitemsnumelMAX_CONSTANT_NUMEL_INLINEtorch._tensor_strr| thresholdreprtorchis_floating_pointlistshapedtype _parameterscode)rrrytab model_str module_nameru module_str example_param buffer_namebufferr| tensor_str param_nameparam maybe_devices r5convertNNModuleToString.converts$6OO  $#4#6 K"OO-.J!!2!2!4d;M(]-B-B *|73 !G9E+c*RH HI$7$&;;#4#4#6 K~||~!::8!++/HHHH!&\ ((00+D,>+?x ~UVW -T&,,-?,@VWX~~ *|73 7)1+c*SQ I#$7*"$!5!5!7 J}L}}0 :4 ;L:MXV[VaVaUbcobpprsJ !G9E*S BG GI"8  277A./r22 r8r\N)rrtorch.fx.GraphModulerRrU)rrrrRrQ) rWrXrYrZrnnLinearConv1dConv2dConv3d BatchNorm1d BatchNorm2d BatchNorm3d LayerNormDropoutSoftmaxReLUGELUIdentity MaxPool2d EmbeddingTanhConvTranspose1dGLULSTMFlattenAdaptiveAvgPool2dro staticmethodrvrr[r\r8r5riris0                            ""+J0  ==r8ric[RR5(dgSn[RRcl[ R "SS/5nUR5RS5nSRUVs/sHo3S:wdM SUS3PM sn5nXS3- nOUS - n[S [R5555nUS - nUR5HupgUSUS US3- nM US- nU$s snf![[ R4a US - nNf=f)Nz:# torch.cuda.is_available()==False, no GPU info collected z# CUDA Info: nvccz --versionrr# z z'# Not searching for nvcc on ROCM setup z# nvcc not found c3`# UH$n[RRU5v M& g7fN)rcudaget_device_name).0is r5 ,_cuda_system_info_comment..s&/O! ""1%%/Os,.z# GPU Hardware Info: z : )rr is_availableversionhip subprocess check_outputdecoder*r:FileNotFoundErrorCalledProcessErrorrrange device_countr)rcuda_version_outcuda_version_linesscomment gpu_namesnamecounts r5_cuda_system_info_commentrsJ :: " " $ $L!I * ==   $)66 7LM !1!8!8!:!@!@!F gg4FR4Fqr'{A3c{4FRSG 9B 'I C CI/4UZZ5L5L5N/OI**I ( r$s5'-- ) I S z<< =*)) *s0A!D" D D D"/D"D""EEF) stable_outputc ^^U(ag/SQm/SQmS UU4Sjjn[RR5VVs/sHup#U"U5(dMSUSUS3PM nnnSRU5n[ S US 35$s snnf) zd Generate a string configuration for environment variables related to Dynamo, Inductor, and Triton. z+# env var omitted due to stable_output=True)TORCHDYNAMOINDUCTORTRITON)TRITON_LIBDEVICE_PATHTRITON_PTXAS_PATHTRITON_LIBCUDA_PATHcH>^[U4SjT55=(a TT;$)Nc3,># UH oT;v M g7frr\)rstringkeys r5r;generate_env_vars_string..filter..)s:zVS=zs)any)r allow_list skip_lists`r5filter(generate_env_vars_string..filter(s:z::Ss)?SSr8z os.environ['z'] = ''rz import os z )rrQrRrU)r(environrr:r )rrrvalue config_lines config_stringrrs @@r5generate_env_vars_stringrs<:JUITT ****,,JC #; +,se6%*, IIl+M #)%  s B Bc hSSKnSSKnU(agURRRR 5nSUR RR 5SURRR 5SURRR 5SUS3 $)Nrz*# config omitted due to stable_output=Truez~import torch._dynamo.config import torch._inductor.config import torch._functorch.config import torch.fx.experimental._config r) torch._functorch.configtorch._inductor.configfx experimental_configcodegen_config_dynamor _inductor _functorch)rrexperimental_configs r5generate_config_stringr7s"!;((//77FFH $$&'(&&()*'')*+  r8cR[RR[5S5$)Nzminifier_launcher.py)r(r)r:rfr\r8r5get_minifier_repro_pathrKs 77<< (> ??r8ch[5n[RSU5 [(a[ U5R 5 [ US5nUR U5 SSS5 g!,(df  g=f![a)n[RS5 [SU35UeSnAff=f)NzWriting minified repro to: %srErzCould not write to ) rrJrKuse_buckr!rHrGOSError exceptionNotImplementedError)contentsminified_repro_pathrNes r5helper_for_dump_minifyrOs13KK02EFx,-335V %s +r HHX , + + V b!$78K7L"MNTUUVs6 A>A-$A>- A;7A>;A>> B1$B,,B1c\rSrSrSrg) AccuracyErrori^r\N)rWrXrYrZr[r\r8r5rr^sr8rc[U5n[[U55HEn[X[R 5(dM&XR XR5 MG U$)z This clone inputs is different from utils clone_input. In case of minifier, all the tensors are leaf tensors while creating a new graph. So, we set the requires_grad field w/o checking the leafness of the tensor. )rrrq isinstancerr requires_grad_ requires_grad)example_inputs cloned_inputsidxs r5clone_inputs_retaining_gradnessr bsX !0MS() m(%,, 7 7   - -n.A.O.O P* r8cXSSKJnJnJn [R "U5nU(d [ U5n[US5(aURS5 [USS5(aU"U5OU"U6nU(aU$U"U5(aU"U5nUR5 U"XSU5$)z Runs a forward and possibly backward iteration for a given mod and args. When disable_clone is True, we will use args as-is without cloning. This is higher fidelity but we may destroy the args in the process. r)collect_resultsreduce_to_scalar_lossrequires_bwd_pass zero_gradT _boxed_callFN) testingrrrcopydeepcopyr hasattrrgetattrbackward) rrargsonly_fwd disable_clonerrroutlosss r5run_fwd_maybe_bwdrosSR r B .t4r; Tb-77"T(RYC $S)  2D$ //r8 require_fp64 ignore_non_fpc SSKJn [XU5nSn[R(a8[ [ R"U5[U55up[XU5n[XU5n U"UU U[RSUS9n U $![a* U(a [S5e[RS5 N[f=f![a [RS5 gf=f) aI Check two models have same accuracy. require_fp64: if True, raise an error if we unable to calculate the fp64 reference ignore_non_fp: if True, do not compare outputs which are not floating point. This is mostly useful for the minifier (which wants to avoid quantizing floating point error into integer/boolean error) r)sameNzfCould not generate fp64 outputs, workaround with torch._dynamo.config.same_two_models_use_fp64 = FalsezCould not generate fp64 outputszWhile minifying the program in accuracy minification mode, ran into a runtime exception which is likely an unrelated issue. Skipping this graph.T)tol equal_nanr!)utilsr#rrsame_two_models_use_fp64 cast_to_fp64rrr  Exception RuntimeErrorrJrKrrepro_tolerance) rropt_gmr rr r!r#reffp64_ref fp64_model fp64_examplesrespassings r5same_two_modelsr3s" B 9CH && ;(4 b!#B>#R) %J)HMH A    " "# G N7 ;"| KK9 :  ;   $ s#7B # C 1B>=B>C#"C#cURRGH?nURS:XaUR[R R RRLa}[UR5S:Xde[URS5(aEURS[R:wa$URS[R4Ul URS:XdMURRS5nUcM[U5(dGM[UR5n[RUS'X1l GMB URR!5 UR#5 U$)N call_functionr{rrr)graphnodesopr/ropsprimsconvert_element_typedefaultrqrr float64kwargsgetdictlint recompile)modelnoder new_kwargss r5cast_dtype_args_to_fp64rFs !! GG & uyyCCKKKtyy>Q& &&diil++ !  0M!YYq\5==9 77o %KKOOG,E ^E%:%:!$++. &+mm 7#( " KK OO Lr8c^SSKJn URT5nT[R:Xa [ U5nU"U4SjU5nX4$)Nr)tree_mapc>[U[R5(a&UR5(aUR T5$U$r)rrr rto)r?rs r5cast_to..s< a & &1+>+>+@+@$$u+ r8)torch.utils._pytreerHrJrr=rF)rrCinputsrHs` r5cast_torOsK- HHUOE  (.    F =r8c6[[RX5$r)rOrr=)rCrNs r5r(r(s 5==% 00r8c U"[R"U5[U55n[UUUUUUS9(+$![a [ R S5 gf=f)NrzWhile minifying the program in accuracy minification mode, ran into a runtime exception which is likely an unrelated issue. Skipping this graphF)rrr r3r)rJr)rrr  compiler_fnrr r! compiled_gms r5backend_accuracy_failsrTst! MM" >~N #    %'       # s8;AAc8UbU$[R"U5$r)r&make_contiguous_strides_for)striders r5_stride_or_defaultrX!s '6UU-N-Nu-UUr8c^U4Sj$)Nc>UbU$T$rr\)r?ds r5rK_mk_defaulter..*s!-Q.Q.r8r\)r[s`r5 _mk_defaulterr])s ..r8cpucX\rSrSrS SjrSSS.S SjjrS SjrS SjrSrg) NopInputReaderi4cSUlg)Nrtotal)r2s r5r6NopInputReader.__init__5s  r8Ndevice dtype_hintc.U=RS- slg)Nrrb)r2 storage_hashnbytesrfrgs r5storageNopInputReader.storage8s a r8cgrr\r2rr>s r5tensorNopInputReader.tensorB r8cgrr\rns r5symintNopInputReader.symintErqr8rbrRrS) ri Optional[str]rjintrf,Optional[torch._prims_common.DeviceLikeType]rgOptional[torch.dtype]rRrS)rrr>rrRzOptional[torch.Tensor])rrr>rrR Optional[int] rWrXrYrZr6rkrorsr[r\r8r5r`r`4sP@D,0 # =  *    r8r`c\rSrSrS SS.S SjjjrSSS.S SjjrS SSSSS.SSjjjrSS jrS rg) InputReaderiKN)pbarcxUc[RS5 Ub [U5OSUl/UlX lg)Nz0no save_dir specified, will generate random data)rJrKrstorerr~)r2save_dirr~s r5r6InputReader.__init__Ls7   KKJ K5=5I'1t !  r8recURbURRS5 [U5n[U5nURbPUbMURR U5nX5R :wa [RSX5R 5 U$[SUS35 X$R-4n[SUS9n[XgXC5R5$![a NNf=f)Nrzdevice mismatch: %s != %szcould not load z , generating random data insteadr)r~update_device_or_default_dtype_or_defaultr read_storagerfrJrKrritemsizerXr untyped_storage)r2rirjrfrgrkrrWs r5rkInputReader.storageYs 99 II  Q #F+&z2 :: !l&> **11,?^^+KK ;V^^TOL>1QRS...0#D6E:>NNPP%  sC C+*C+)storage_offsetrris_leafc ,[X2S9n[U5n[U5n[U5n[ U5n[ R "/XQRUS9n [ R"5 U RXX#5 SSS5 U(dk[ R"5 U R[ RS9n SSS5 [ R"5 U RXX#5 SSS5 [ RRRU 5U:Xde[ R R#X5 UR$R'U 5 U $!,(df  N=f!,(df  N=f!,(df  N=f)Nr)rrfr) memory_format)rX_storage_offset_or_defaultr_is_leaf_or_default_requires_grad_or_defaultrrorfno_gradset_ enable_gradclonepreserve_format _subclasses meta_utils safe_is_leaf_utilsset_tensor_metadatarappend) r2rkrrWrrrrmetadatats r5roInputReader.tensorvs$F83NC!%(%g.1-@ LL eNN- ]]_ FF7E :""$GG%*?*?G@%w>!  ++88;wFFF ((5 _%$ s$+E#"E4F# E14 F Fc<URRU5 U$r)rr)r2vals r5rsInputReader.symints  r8)rr~rr)rz str | Noner~z tqdm | NonerRrS) rirvrjrwrfrxrgryrRr)rkrrtorch._prims_common.ShapeTyperW(Optional[torch._prims_common.StrideType]rrzrryrOptional[bool]rrrrrR torch.Tensor)rrrRrr{r\r8r5r}r}Ks%) BF " 4?   $@D,0 Q#QQ = Q * Q QB<@  )-'+(,"&-9  & %&  Br8r}cz\rSrSrSS.SSjjrSSjrSSS.SSjjrSS jrSS jrSS jr SS jr S r g) InputWriteriF stable_hashc/Ul[R"5UlXlUb [ XS9OSUl0Ulg)Nr)_lines itertoolsrstorage_counterrrr seen_storages)r2rrs r5r6InputWriter.__init__sC!# (0 # x A 9;r8crS/nURSUR55 URS5 U$)Nzdef load_args(reader):c3,# UH nSU3v M g7f)rzNr\)rls r5r$InputWriter.lines..s1[4s[szload_args._version = 0)extendrr)r2rs r5linesInputWriter.liness8 $  1T[[11 )*r8N) device_hintrgc X[U5nURRU5nUbU$S[UR53nSn[ S5[ U5:waSU<3nSnUR nURS:XaUceUn[S5U:waSU<3nUR5n Sn URb5UR RS:waURRU5n URRUSU <SU <UUS35 XPRU'U$) Nbufrz , dtype_hint=metaz , device=z = reader.storage(, r})rrr?rrrrfrnrrjr write_storagerr) r2rrrgwsvmaybe_dtype_hintrrfrjris r5rkInputWriter.storagesBO ,    " "2 & =H$t++,- . T "&7 &C C!.zn=   '' ;;& * ** F d #v -&vj1L '') :: !o&<&<&A&AV&K::33ODL c#L#32fZ ~N^M__` a "#2r8c SSKJnJn URUR 5UR UR S9n/nU"U"[SURS9UR555(d1UR[[UR5555 [S5UR :waURSUR <35 U"[S5UR5:H5(d#URSUR5<35 [ R"R%U5nU(a&UR'SUR)555 [+S5UR,:waURSUR,<35 [ R.R0R3U5n[5S5U:waURS U<35 UR6RS S R9U[[UR55/UQ5-S U3-5 g) Nr)statically_known_truesym_eq)rgrrzdtype=zstorage_offset=c36# UHupUSU<3v M g7f)=Nr\)rkrs r5r%InputWriter.tensor..sI1H1#Qqe 1Hszrequires_grad=zis_leaf=zreader.tensor(r) # )%torch.fx.experimental.symbolic_shapesrrrkrrrfrXrrWrrQtuplerrrrrget_tensor_metadatarrrrrrrrrr:) r2rrrrrkrtensor_metadatars r5roInputWriter.tensorsW,,   AGG $ %d!'':AHHJ G   KKE!((*-. / T "agg - KK& , -$ &t ,0@0@0B B   KK/!*:*:*<)?@ A,,::1=  KKI1F1F1HI I $T *aoo = KK.(;< =##..;;A> t $ / KK(7+. /  ii#eAGGn"5==> ?dVn  r8c*URRSUS[U535 [U[[ 45(aURRS5 [ U5Hup4USUS3n[U[R5(aURXT5 M?[U[[R45(aURXT5 MwURXT5 M URRS5 gg)Nrz# was unsupported type for dumping: z"""[])rrrnrrr enumeraterr rorwSymIntrs unsupported)r2rargraname_is r5rInputWriter.unsupporteds Rv%Hc TU cD%= ) ) KK  u %!# 61#Qa..KK*C#677KK*$$V/' KK  u % *r8cJURRSU<SUS35 g)Nz reader.const(rz!, filtered out during compilation)rr)r2rs r5constInputWriter.consts' D85.O P r8c[U[R5(aURRnUR R SU<SU35 g)Nzreader.symint(r)rrrrDhintrr)r2rrs r5rsInputWriter.symints? c5<< ( (((--C ^C7%v>?r8)rrrrr)rrvrrUrRrS)rRrV)rrrrxrgryrRrQ)rrQrrrRrS)rrQrrrRrS)rrQrRrS)rrQrrrRrS) rWrXrYrZr6rrkrorrrsr[r\r8r5rrs\GL ;&EI,0 ' B  *  D @&$ @r8rc ^^^^SSKJn UR5VVs0sHupVXe_M nnnSRUR 55n[ R "U5n SUS3n SUS3n Sn "S S 5n 0nU=(d 0mSUU4S jjmSUU4S jjnURnUR"5HunnUS :XaM[R"U U5nU(a>UR5unn[URS55nUUnU"UU5UU'[R"U U5nU(dMT"URS55UU'M S[ R"U5R;aqU "5nUUS'[R "X5HKnUR5unnnn[URS55nUUn[#UUU"UU55 MM U$s snnf)a Takes in a function which has been printed with print_readable() and constructs kwargs to run it. Handles Tensor inputs, Symints, and a graph module which might have tensor constants. Consider a function `forward` defined as follows: def forward(self, primals_1: "f32[1001, 6]", primals_2: "f32[s0]", primals_3: "Sym(s0)",): _tensor_constant0: "i64[4190]" = self._tensor_constant0 # Further implementation kwargs = aot_graph_input_parser(forward) forward(**kwargs) r) dtype_abbrs|z(_tensor_constant\d+): \"(z0)\[\s*(.*?)\s*\]\" = self\.(_tensor_constant\d+)(z)\[\s*(.*?)\s*\]z Sym\((s\d+)\)c\rSrSrSrSrg)/aot_graph_input_parser..TensorContaineriHz#Container for tensors as attributesr\N)rWrXrYrZ__doc__r[r\r8r5TensorContainerrHs-r8rrRc|>^[R"TT;=(d TSLU4Sj5 TRTT5$)Nc>TS3$)Nz; not in symbolic_shapes and default sym shape not passed inr\)rssr5rK=aot_graph_input_parser..get_sym_int..Ss vhYZr8)r_checkr?)rsdefault_sym_shapesym_shapes_dicts`r5 get_sym_int+aot_graph_input_parser..get_sym_intPs; o % F):$)F Z ""6+<==r8c>/n/n[U5HjupEUR5nSU;a,T "U5nURU5 URU5 MGU(dMPUR[U55 Ml UR(a[ R O[ RnU"X!T S9nUH"n [ RRX5 M$ U$)Nr)rrf) rstriprrwrrrandnzerosr mark_dynamic) rrresolved_shape dynamic_dimsrdimr constructorrr[rfrs r5 gen_tensor*aot_graph_input_parser..gen_tensorWs &FA))+Ccz$%%a(##A&3"))#c(3'&+%<%rrr annotationmatch data_type shape_strrr container attr_namertrrs ` ` @@r5aot_graph_input_parserr$s*5&1%6%6%8)%8zs %8)+"4"4"67M  t $F">*&&K(..0z H   , 3 #(<<> Iy)//#./Ei(E&ue4F5M /:6 5' A7F5M1"""4(333#% "v[[!8AE16 .Iy)Q)//#./Ei(E Iy*UE*B C B M[)sG7c^^[R"5m[RR [RR T55mSU4SjjnSUU4Sjjn[ R"U5 U$)z Decorator to cProfile a given function and save the result to disk on process exit. Args: filename: filename to save profile to cN>^[R"T5SUU4Sjj5nU$)Nc>TR5 T"U0UD6TR5 $!TR5 f=fr)enabledisable)rr>fnprofs r5wrapper3profile_to_file..decorator..wrappers1 KKM 4*6*  s+=)rrr>rrRr) functoolswraps)rr rs` r5 decorator"profile_to_file..decorators%     r8c >TRT5 [RR[R "STSTS355 g)Nz! Wrote profile to z+, view with: snakeviz z ) dump_statssysstderrrHr<r=)r3rsr5save_it profile_to_file..save_itsK !  OO""*,&Z(  r8)rrrRrru)cProfileProfiler(r)r+ expanduseratexitregister)r3r$r*rs` @r5profile_to_filer1sW    Dwwrww11(;py build_info BuildInfoget_build_ruler.r:rIr!rfrricacherrrrrr)rr rr3rFrOr(rTrXr]float32rrfrrrrr`r}rrr1)r?s0r5rPs$#   #88 ##.@.;M992,! CL 89  $ $ &     J $$..==?GG SWXJIITA!9!B?TUM644nccL 67< 25:(@ V I    00 00 0  0H 7 77 7"7 7  77 7t* 3=F+(1 1)21+1  !S    RV 4V )V$ V/"%--0"5<<#67*1-)%0#E*  .NNzy@y@|+/'+ d 0d d)d% d  dN#qUsJ5