+ j\(Rt^RIHt^RIt^RIt^RIt^RIt^RIt^RIt^RI t ^RI t ^RI t ^RI t ^RI t ^RIt^RIt^RIt^RIHt^RIHt^RIHtHtHtHt^RIt^RIHt^RIt^RIHt^RI H!t!^RI"H#t#^R IH$t$^R I%H&t&^R I'H(t(^R I)H*t*H+t+^R I,H-t-^RIH.t.H/t/H0t0]'d^RI1H2t2H3t3^RI4H5t5^RI6H7t7] Pp!]94t:]!R4t;]!R4t<]]>'d^RI?t@.tARtBRtC]>'da.RROtA]@PPPP4PRR4tCRPR]A44tJR]J R2tB.RSOtK!RR4tLRRltM^tN!RR ]O4tP!R!R"4tQ]PR#R$l4tSR%R&/R'R(lltTR%R&/R)R*lltUR+R,ltVR-R.ltW!R/R0]X4tYR1R2ltZRTR3R4llt[RUR5R&R6R&/R7R8lllt\R9R:lt]R;R<lt^R=R>lt_RUR5R&R6R&/R?R@lllt`RARBltaRCRDltb]b!]P4td]b!]P!RE44tf]b!^4tg]b!R&4th]b!R&4ti!RFRG4tj!RHRI4tkRJRKltl!RLRM4tmRVRNROlltnRPRQltoR#)Wa Debug utilities for TorchDynamo compilation and execution. This module provides various debugging tools and utilities for TorchDynamo, including: - Minification support for reducing test cases while preserving bugs - Input/output handling via InputReader and InputWriter for reproducible testing - Accuracy checking between original and compiled models - Neural network module string conversion via NNModuleToString - Profiling tools and system information collection - Buck build system integration for Meta-internal testing Key classes: - InputReader/InputWriter: Handle serialization of model inputs/outputs - NNModuleToString: Converts nn.Modules to string representations - BuckTargetWriter: Manages Buck build system integration ) annotationsN)Counter) import_module)Anycast TYPE_CHECKINGTypeVar)Tensor) rand_strided)normalize_path_separator)is_float_dtype)StorageWeakRef) ConfigModule)ContentStoreReaderContentStoreWriter)config) clone_inputs get_debug_dir warn_once)CallableSequence)tqdm)UntypedStorageTztorch._inductor.configzfbcode:// c#."TF pRV R2xK R#5i)z "",N).0xs& Z/home/luca/screen-checking/.venv/lib/python3.14/site-packages/torch/_dynamo/debug_utils.py r#Ys C 152 zfor _extra_dep in [ zY ]: try: torch.ops.load_library(_extra_dep) except OSError: pass c>]tRt^ftRRltRRltR RRlltRtR#) BuckTargetWriterc V^8dQhRRRR/#)filenamestrreturnNoner)formats"r" __annotate__BuckTargetWriter.__annotate__gs 6 6 6 6c >\PP\PPV44wVnVnVP P RR4VnVPP RR4 RVP 2VnVPVPPR4RVnVPR,VnVPpW"PR4RR,pRV R VP 2Vn R#) z.pyr/.zfbcode.N:NNzfbcode/r:) ospathsplitabspathsubdirpy_filereplacetargetfind cmd_line_path)selfr)tmps&& r"__init__BuckTargetWriter.__init__gs$&GGMM"''//(2K$L! T\ll**5"5 {{**345Qt{{mD IIdiinnY79: IIbM kk((9%'(,!#a }5r0cV^8dQhRR/#r(r+r*r)r-s"r"r.r/us  s r0c  RP\Uu.uF pRV R2NK up4p\P!RVP RVP R\ RV RVP R2 4#uupi) rz "rza 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_targetr7)r@r!extra_cpp_depss& r"buildBuckTargetWriter.buildusz#Jz!is"$5z#JK  ;;-ll^  II;#   $KsA2c V^8dQhRRRR/#)r( print_msgboolr+ list[str]r)r-s"r"r.r/s  t y r0c \PPVPR4p\ VR4;_uu_4pVP VP 44RRR4\VP.,pV'd&\PRRPV44V# +'giLW;i)TARGETSwNzFFound an example that reproduces the error. Run this cmd to repro - %s ) r6r7rGr:openwriterMBUCK_CMD_PREFIXr?logwarning)r@rP target_filefd cmd_splits&& r"rXBuckTargetWriter.writesggll4;; : +s # #r HHTZZ\ "$$t'9'9&::  KKX# $ #s B.. B> )r?r7r;r:r=N)T)__name__ __module__ __qualname____firstlineno__rBrMrX__static_attributes__rr0r"r&r&fs 6 4  r0r&cV^8dQhRR/#rEr)r-s"r"r.r.scr0c2\PP\4R4pVf.\P !4 R\ P!4 2p\PPV4'g\P!VRR7V#)minifierz /minifier_T)exist_ok) r6r7rGrtempfile gettempdirgetpassgetuserexistsmakedirs)r7s r" minifier_dirrose 77<<  4D |%%'( 7??3D2EF 77>>$   D4( Kr0c]tRt^tRtR#)UnsupportedNNModuleErrorrNr`rarbrcrdrr0r"rqrqr0rqc]tRt^tRR0t0RmtRR0tRt]PP]PP]PP]PP]PP]PP]PP ]PP"]PP$]PP&]PP(]PP*]PP,]PP.]PP0]PP2]PP4]PP6]PP8]PP:]PP<.t] RRl4t!] RRl4t"] R R l4t#] R R l4t$] R Rl4t%] RRl4t&] RRl4t'] RRl4t(] RR/RRll4t)] RR/RRll4t*Rt+R#)NNModuleToStringz0torch.ao.quantization.fake_quantize.FakeQuantizezAtorch.ao.quantization.fake_quantize.FusedMovingAvgObsFakeQuantizez%torch.ao.nn.qat.modules.linear.Linearz8torch.ao.nn.intrinsic.qat.modules.linear_relu.LinearReLUc V^8dQhRRRR/#)r(gmtorch.fx.GraphModuler+rQr)r-s"r"r.NNModuleToString.__annotate__s"64r0c H\4pVP4F[wr#\V\PP 4'dK1\ PV4eKJVPV4K] \V4^8d\PRV4R#R#)Nz2Cannot safely convert some modules to strings - %sFT) setnamed_children isinstancetorchfx GraphModuleru_module_constructor_stringaddlenrZr[)rw cant_convert_modules& r"can_convert_to_string&NNModuleToString.can_convert_to_stringsu **,IA&%(("6"677::6BJ  ( - | q KKDl r0c V^8dQhRRRR/#r(rtorch.nn.Moduler+r*r)r-s"r"r.rysDD/DcDr0c \\V4P R\V4P 2#r3)typerar`)rs&r"_module_type_name"NNModuleToString._module_type_names)v,))*!DL,A,A+BCCr0c V^8dQhRRRR/#)r(type_z type[object]r+r*r)r-s"r"r.rys66,636r0c 8VP RVP 2#r)rar`)rs&r" _type_nameNNModuleToString._type_names""#1U^^$455r0c V^8dQhRRRR/#r(rrr+ str | Noner)r-s"r"r.rys333J3r0c \PV4pV\P9dR#VPp\ \ VP 4p\ \ VP4p\VRR4'dV^,pV^,^,pR\P\V44 2RV: 2RV: 2RVP: 2RVP: 2RVP: 2.p\VR4'd VPR VP : 24\VR 4'd VPR VP": 24V R R P%V4 R2#)N reduce_rangeFz observer=z quant_min=z quant_max=dtype=zqscheme=z is_dynamic=z reduce_range=ch_axiszch_axis=(, ))rurfake_quant_modulesactivation_post_processrint quant_min quant_maxgetattrrrdtypeqscheme is_dynamichasattrappendrrrG)r module_typeobserverrrargss& r"_fake_quant_constructor(NNModuleToString._fake_quant_constructorsJ&88@ .AA A11f../ f../ 8^U 3 3 NI!A )I(33DNCD E & &V\\$ %v~~( )&++. /   8^ , , KK-(=(='@A B 8Y ' ' KK(8#3#3"67 8a $022r0c V^8dQhRRRR/#rr)r-s"r"r.rys   Z r0c \VRR4p\PFOp\PP P V4p\T4\T48XgKHRT: R2u# R# \dKci;i)qconfigNz6qconfig=torch.ao.quantization.get_default_qat_qconfig(r) rrudefault_qat_qconfig_backendsr~ao quantizationget_default_qat_qconfigAssertionErrorrepr)rrbackenddefault_qconfigs& r"_qat_qconfig_constructor)NNModuleToString._qat_qconfig_constructors&)T2'DDG "'(("7"7"O"OPW"XG}_ 55{!% E"  s)A33 BBc V^8dQhRRRR/#rr)r-s"r"r.ry's33o3*3r0c  \PV4pV\P9dR#\PV4pVfR#\ VP 4\ VP 4RVP: 2RVP: 2RVP: 2RVP: 2RVP: 2RVPRJ 2RVP: 2. p\VRR4pVe?VPR VP : 2R VP": 2R VP$: 2.4VP'V4V R R P)V4 R2#)Nz kernel_size=zstride=zpadding=z dilation=zgroups=bias=z padding_mode=bnzeps=z momentum=z freeze_bn=rrr)rurqat_conv_modulesrr in_channels out_channels kernel_sizestridepaddingdilationgroupsbias padding_moderextendepsmomentum freeze_bnrrG)rrrrrs& r"_qat_conv_constructor&NNModuleToString._qat_conv_constructor&sP&88@ .?? ?";;FC ? ## $ $$ %6--0 1fmm& 'v~~( )* +fmm& 'FKKt+, -F//2 3  VT4 ( > KK266*% / !1!1 45  Ga $022r0c V^8dQhRRRR/#rr)r-s"r"r.ryGs333J3r0c 4\PV4pV\P9dR#\PV4pVfR#\ VP 4\ VP 4RVPRJ 2V.pV RRPV4 R2#)Nrrrr) rurqat_linear_modulesrr in_features out_featuresrrG)rrrrs& r"_qat_linear_constructor(NNModuleToString._qat_linear_constructorFs&88@ .AA A";;FC ? ## $ $$ %FKKt+, -    a $022r0c$V^8dQhRRRRRR/#)r(rr module_strr*r+rQr)r-s"r"r.ryXs"_#RVr0c \V4\P9dR#\RV 2RR4R# \dR#i;i)Fzmod = z execT)rru safe_reprscompile SyntaxError)rrs&&r"_can_emit_module_constructor-NNModuleToString._can_emit_module_constructorWsI </:: :  fZL)?F C  s4 AAallow_unsafe_reprFc$V^8dQhRRRRRR/#)r(rrrrQr+rr)r-s"r"r.rycs$7; r0c \V4p\PW4'dV#\PV4;'g3\P V4;'g\P V4pVeV#V'dV#R#N)rrurrrr)rrrconstructor_strs&$ r"r+NNModuleToString._module_constructor_stringbs&\  8 8 L L   4 4V < @ @55f= @ @77?   &" "  r0c$V^8dQhRRRRRR/#)r(rwrxrrQr+r*r)r-s"r"r.ryus-]W]W(]W]WQT]Wr0c aaaaaaa^RIHoR o/o.oRVVV3RlloRVVVV3RlloRS!V4 2pRRPSV.,4,#) ) _addindentc$V^8dQhRRRRRR/#)r( name_hintr*subrxr+r)r-s"r"r..NNModuleToString.convert..__annotate__s"   3G C r0c<\V4S9dS\V4,#V R\S4 2pVS\V4&S!V4pSPRV RV 24V#)rzclass z(torch.nn.Module): )idrr)rr class_namebodyemit_module_bodysubgraph_classes subgraph_defss&& r"register_subgraph3NNModuleToString.convert..register_subgraphss#w**'300%;a,<(='>?J(2 RW %#C(D  6*5I$!P Q r0c V^8dQhRRRR/#rr)r-s"r"r.rs@ @ _@ @ r0c@<S RS^, R2pVP4EFwr#\V\PP4'd$S!W#4pVS^, RV RV R2, pKS\ P VSR7pVf\RV: 24h\VP4R4pVeTM\VP4R4pVe.VPPR8wdV R VP R 2pVS^, RV RV R 2, pEK VPP4EFwrV fK V P4\ 8:dH^R IHp V P&\ 8d!\)R V P& R\ R24h\+V 4p Mi\P,!V 4'd(R\/V P04 RV P2 R2p M&R\/V P04 RV P2 R2p V PPR8wdV R V P R 2p VS^, RV RV R2, pEK VP4P4FywrV fK RpV PPR8wdRV P R2pR\/V P04 RV P2 V R2p VS^, RV RV R 2, pK{ VS!VP6\9S44 R 2, pV#)zdef __init__(self) -> None: zsuper().__init__() zself.z = z() )rNz!Cannot convert module to string: cpuz.to("z")r) PRINT_OPTSzPRINT_OPTS.threshold (z() must be >= MAX_CONSTANT_NUMEL_INLINE (rz torch.randn(z, dtype=ztorch.randint(1, size=zself.register_buffer('z', z) rz , device=""ztorch.nn.Parameter(torch.randn(z)))r|r}r~rrrurrqnext parametersbuffersdevicer_buffersitemsnumelMAX_CONSTANT_NUMEL_INLINEtorch._tensor_strr thresholdrris_floating_pointlistshaper _parameterscoder)r model_str module_namechildrr example_paramexample_tensor buffer_namebufferr tensor_str param_nameparam maybe_devicerrrtabs& r"r2NNModuleToString.convert..emit_module_bodys;%4S1WI=QR '-&;&;&=" eUXX%9%9::!2;!FJC!G9E+c*T!RRI-HH->I %2;E9E !%U%5%5%7 > %0"emmot4 "-.2G2G2L2LPU2U$.0(.'<'<'># ><<>%>><!++.GG,4Z5I5I4JK::S9TTUW"&fJ,,V44&tFLL'9&:(6<<.PQR$:$v||:L9MXV\VbVbUccd!eJ==%%.$.tEKK?P>QQYZ_ZeZeYfgsfttvw ayj\ZLKK &@ Jv{{CH=>bA AI r0zclass Repro(torch.nn.Module): z from torch.nn import * r )torch.nn.modules.modulerrG) rwr repro_classrrrrrrs &d @@@@@@r"convertNNModuleToString.convertts_6,.#%   @ @ D88H8L7MN +dii  8U.VVVr0rN> #torch.ao.nn.qat.modules.conv.Conv1d#torch.ao.nn.qat.modules.conv.Conv2d#torch.ao.nn.qat.modules.conv.Conv3d5torch.ao.nn.intrinsic.qat.modules.conv_fused.ConvBn1d5torch.ao.nn.intrinsic.qat.modules.conv_fused.ConvBn2d5torch.ao.nn.intrinsic.qat.modules.conv_fused.ConvBn3d7torch.ao.nn.intrinsic.qat.modules.conv_fused.ConvReLU1d7torch.ao.nn.intrinsic.qat.modules.conv_fused.ConvReLU2d7torch.ao.nn.intrinsic.qat.modules.conv_fused.ConvReLU3d9torch.ao.nn.intrinsic.qat.modules.conv_fused.ConvBnReLU1d9torch.ao.nn.intrinsic.qat.modules.conv_fused.ConvBnReLU2d9torch.ao.nn.intrinsic.qat.modules.conv_fused.ConvBnReLU3d)fbgemmx86qnnpackonednn),r`rarbrcrrrrr~nnLinearConv1dConv2dConv3d BatchNorm1d BatchNorm2d BatchNorm3d LayerNormDropoutSoftmaxReLUGELUIdentity MaxPool2d EmbeddingTanhConvTranspose1dGLULSTMFlattenAdaptiveAvgPool2dr staticmethodrrrrrrrrrrrdrr0r"rurus:K  0B$J                             ""+J0"DD66332  33>33 >C"]Wu]W]Wr0rucV^8dQhRR/#rEr)r-s"r"r.r.s3r0c\PP4'gR#Rp\PPfq\ P !RR.4pVP4PR4pRPVUu.uFq3R8wgK RV R2NK up4pW R2, pM VR, p\R \\PP4444pVR , pVP4FwrgVRV R V R2, pK VR, pV#uupi \\ P3d TR , pLi;i) z:# torch.cuda.is_available()==False, no GPU info collected z# CUDA Info: nvccz --versionrrz# z z'# Not searching for nvcc on ROCM setup z# nvcc not found c3`"TF$p\PPV4xK& R#5ir)r~cudaget_device_name)r is& r"r#,_cuda_system_info_comment..s&/O! ""1%%/Os,.z# GPU Hardware Info: z : )r~rI is_availableversionhip subprocess check_outputdecoder8rGOSErrorCalledProcessErrorrrange device_countr)r cuda_version_outcuda_version_linesscomment gpu_namesnamecounts r"_cuda_system_info_commentr^sJ :: " " $ $L!I * ==   $)66 7LM !1!8!8!:!@!@!F gg4FR4Fqr'{A3c{4FRSG 9B 'I C CI/4UZZ5L5L5N/OI**I ( r$s5'-- ) I S Z22 3*)) *s0A"D9 D4 D4#D96 D94D99$E E  stable_outputFc V^8dQhRRRR/#r(r_rQr+r*r)r-s"r"r.r.s  t  r0c vaaaV'dR#.ROo.ROoRR.oRVVV3Rllp\PP4UUu.uF.wr#V!V4'gKRV RVPRR 4 R2NK0 pppVP R S44R P V4p\ R V R 24#uuppi)zd Generate a string configuration for environment variables related to Dynamo, Inductor, and Triton. z+# env var omitted due to stable_output=TrueTORCHDYNAMO_REPRO_AFTERTORCHDYNAMO_REPRO_LEVELc V^8dQhRRRR/#)r(keyr*r+rQr)r-s"r"r..generate_env_vars_string..__annotate__s  C D r0c<a\;QJdV3RlS4F 'gK RM RM!V3RlS44;'dSS9;'dSS9#)c3,<"TF qS9xK R#5irr)r stringrfs& r"r#;generate_env_vars_string..filter..s7J&# JsTF)any)rf allow_listrepro_env_vars skip_listsfr"filter(generate_env_vars_string..filtersJ C7J7CCC7J7 7 * *9$ * *>) r0z os.environ['z'] = ''rc3."TF pRV R2xK R#5i)zos.environ.pop('z', None)Nr)r rfs& r"r#+generate_env_vars_string.. sSNS*3%x8Nr$rz import os z )TORCHDYNAMOINDUCTORTRITON)TRITON_LIBDEVICE_PATHTRITON_PTXAS_PATHTRITON_LIBCUDA_PATH)r6environrr<rrGr ) r_rprfvalue config_lines config_stringrmrnros $ @@@r"generate_env_vars_stringrs<:JUI/1JKN  ****,,JC #; ALVEMM#s$;# ??r0c V^8dQhRRRR/#)r(contentsr*r+r,r)r-s"r"r.r.7s V VS VT Vr0c\4p\PRV4\'d\ V4P 4\ VR4;_uu_4pVP V4RRR4R# +'giR#;i \d*p\PR4\RT 24ThRp?ii;i)zWriting minified repro to: %srUNrzCould not write to ) rrZr[use_buckr&rXrWrS exceptionNotImplementedError)rminified_repro_pathr]es& r"helper_for_dump_minifyr7s13KK02EFx,-335V %s + +r HHX , + + + V b!$78K7L"MNTUUVs6B A7-B 7 B B B B?$B::B?c]tRtRtRtR#) AccuracyErroriFrNrrrr0r"rrFrsr0rc V^8dQhRRRR/#)r(example_inputs Sequence[Any]r+ list[Any]r)r-s"r"r.r.Js  M i r0c\V4p\\V44FRp\W,\P 4'gK+W,P W,P4KT V#)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. )rrUrr}r~r requires_grad_ requires_grad)r cloned_inputsidxs& r"clone_inputs_retaining_gradnessrJsX !0MS() m(%,, 7 7   - -n.A.O.O P* r0c ,V^8dQhRRRRRRRRRR /#) r(rwrxrronly_fwdrQ disable_cloner+rr)r-s"r"r.r.Ws:000 00 0  0r0cd^RIHpHpHp\P !V4pV'g \ V4p\VR4'dVPR4\VRR4'd V!V4MV!V!pV'dV#V!V4'dV!V4pVP4V!WRV4#)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. )collect_resultsreduce_to_scalar_lossrequires_bwd_pass zero_gradT _boxed_callFN) testingrrrcopydeepcopyrrrrbackward) rwrrrrrroutlosss &&&& r"run_fwd_maybe_bwdrWsSR r B .t4r; Tb-77"T(RYC $S)  2D$ //r0 require_fp64 ignore_non_fpc4V^8dQhRRRRRRRRRRR RR R/#) r(rwrxopt_gmrrrrQrrr+r)r-s"r"r.r.wsN777 7"7 7  77 7r0c ^RIHp\WV4pRp\P'd9\ \ P!V4\V44wr\WV4p\WV4p T!TT T\PRTR7p T # \d,T'd \R4h\PR4L`i;i \d\PR4R#i;i)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) )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)utilsrrrsame_two_models_use_fp64 cast_to_fp64rrr Exception RuntimeErrorrZr[rrepro_tolerance) rwrrrrrrreffp64_ref fp64_model fp64_examplesrespassings &&&&$$ r"same_two_modelsrws" B 9CH &&& ;(4 b!#B>#R) %J)HMH A    " "# G N7 ;"| KK9 :  ;   $ s)7B$ CC! CC C+*C+c V^8dQhRRRR/#)r(modelrxr+r)r-s"r"r.r.s#7cast_to..s< a & &1+>+>+@+@$$u+ r0)torch.utils._pytreerrr~rr)rrrrsf&& r"cast_torsK- HHUOE  (.    F =r0c$V^8dQhRRRRRR/#)r(rrxrrr+rr)r-s"r"r.r.s$11 1)21+1r0c6\\PW4#r)rr~r)rrs&&r"rrs 5==% 00r0c4V^8dQhRRRRRRRRR RR RR R/#) r(rwrxrr compiler_fnzACallable[[torch.fx.GraphModule, list[Any]], torch.fx.GraphModule]rrQrrr+r)r-s"r"r.r.sO!S    r0c V!\P!V4\V44p\VVVVVVR7'*# \d\ P R4R#i;i))rrzWhile minifying the program in accuracy minification mode, ran into a runtime exception which is likely an unrelated issue. Skipping this graphF)rrrrrrZr)rwrrrrr compiled_gms&&&&$$ r"backend_accuracy_failsrst! MM" >~N #    %'       # s;> A"!A"c$V^8dQhRRRRRR/#)r(r%torch._prims_common.StrideType | Noner torch._prims_common.ShapeTyper+ztorch._prims_common.StrideTyper)r-s"r"r.r. s-VV 1V )V$ Vr0c:VeV#\P!V4#r)rmake_contiguous_strides_for)rr s&$r"_stride_or_defaultr s '6UU-N-Nu-UUr0c V^8dQhRRRR/#)r(drr+zCallable[[T | None], T]r)r-s"r"r.r.s//Q/2/r0caV3Rl#)c<VeV#S#rr)r!rs&r"r_mk_defaulter..s!-Q.Q.r0r)rsfr" _mk_defaulterrs ..r0rc]tRtRtRRltRRRR/RRlltR R ltR R ltR RltRRlt RRlt RRlt Rt R#)NopInputReadericV^8dQhRR/#r(r+r,r)r-s"r"r.NopInputReader.__annotate__ s$r0c ^VnR#)rNtotal)r@s&r"rBNopInputReader.__init__ s  r0rN dtype_hintc ,V^8dQhRRRRRRRRR R /#) r( storage_hashrnbytesrr)torch._prims_common.DeviceLikeType | Noner torch.dtype | Noner+r,r)r-s"r"r.r#s<  :  '  r0c 8V;P^, unR#Nr )r@rrrr s&&&$$r"storageNopInputReader.storage#s a r0c$V^8dQhRRRRRR/#)r(rrrr+ztorch.Tensor | Noner)r-s"r"r.r-s"  C 3 3F r0c R#rrr@rrs&*,r"tensorNopInputReader.tensor- r0c$V^8dQhRRRRRR/#)r(rrrr+ int | Noner)r-s"r"r.r0s!  C 3 : r0c R#rrrs&*,r"symintNopInputReader.symint0rr0c V^8dQhRRRR/#r(r\r*r+r,r)r-s"r"r.r3s  # $ r0c R#rrr@r\s&&r"constNopInputReader.const3rr0c V^8dQhRRRR/#r$r)r-s"r"r.r6s    r0c R#rrr&s&&r" unsupportedNopInputReader.unsupported6rr0c$V^8dQhRRRRRR/#)r( device_typer* device_indexrr+r,r)r-s"r"r.r9s!  S   r0c R#rr)r@r.r/s&&&r" generatorNopInputReader.generator9rr0c V^8dQhRRRR/#r(script_class_namer*r+r,r)r-s"r"r.r<sr0c 8V;P^, unR#rr r@r5s&&r"opaqueNopInputReader.opaque<s a r0r  r`rarbrcrBrrr!r'r+r1r8rdrr0r"rrsD =A  *.      r0rc ]tRtRtRRR/RRllltRRRR/RR lltRR RR RR RR R/RRllltRR/RRlltRRltRRlt RRlt RRlt Rt R#) InputReaderiBNpbarc$V^8dQhRRRRRR/#)r(save_dirrr=z tqdm | Noner+r,r)r-s"r"r.InputReader.__annotate__Cs$  " 4?   r0c ~Vf\PR4Ve \V4MRVn.VnW nR#)Nz0no save_dir specified, will generate random data)rZr[rstorerr=)r@r?r=s&&$r"rBInputReader.__init__Cs7   KKJ K5=5I'1t !  r0rr c ,V^8dQhRRRRRRRRR R /#) r(rrrrrrr rr+rr)r-s"r"r.r@PsCQQ QQ : Q ' Q Qr0c VPeVPP^4\V4p\V4pVPeSVeOVPP V4pW5P 8wd!\PRW5P 4V#\RV R24W$P,3p\RVR7p\WgWC4P4# \dLVi;i)Nzdevice mismatch: %s != %szcould not load z , generating random data insteadr )r=update_device_or_default_dtype_or_defaultrB read_storagerrZr[FileNotFoundErrorritemsizerr untyped_storage)r@rrrr rr rs&&&$$ r"rInputReader.storagePs 99 II  Q #F+&z2 :: !l&> **11,?^^+KK ;V^^TOL>1QRS...0#D6E:>NNPP%  sC)) C76C7storage_offsetrris_leafc<V^8dQhRRRRRRRRR R R R R R RRRR/ #)r(rrr rrrrOrrrrz bool | NonerPmetadatarr+ torch.Tensorr)r-s"r"r.r@msd###-#6 # # #"##### #r0c \W2R7p\V4p\V4p\V4p\ V4p\ P !.WQPVR7p \ P!4;_uu_4V PWW#4RRR4V'g\ P!4;_uu_4V P\ PR7p RRR4\ P!4;_uu_4V PWW#4RRR4\ PPPV 4V8wd9\!R\ PPPV 4 RV 24h\ P"P%W4VP&P)V 4V # +'giEL9;i +'giL;i +'giL;i)rF)rrrN) memory_formatz/Tensor leaf status mismatch: safe_is_leaf(t) = z , expected )r_storage_offset_or_defaultrI_is_leaf_or_default_requires_grad_or_defaultr~rrno_gradset_ enable_gradclonepreserve_format _subclasses meta_utils safe_is_leafr_utilsset_tensor_metadatarr) r@rr rrOrrrPrRts &&&&$$$$, r"rInputReader.tensormsV$F83NC!%(%g.1-@ LL eNN- ]]__ FF7E :""$$GG%*?*?G@%w>!    ' ' 4 4Q 77 B A$$//<&%,,''H'' (8(899 <r0c]tRtRtRR/RRlltRRltRR R R /R R lltR RltRRltRRlt RRlt RRlt RRlt Rt R #) InputWriteri stable_hashFc$V^8dQhRRRRRR/#)r(r?rrrQr+r,r)r-s"r"r.InputWriter.__annotate__s! ; ; ;T ;d ;r0c .Vn\P!4VnWnVe \ WR7MRVn/VnR#)N)r)_lines itertoolsr]storage_counterr?rrB seen_storages)r@r?rs&&$r"rBInputWriter.__init__sC!# (0 # x A 9;r0cV^8dQhRR/#)r(r+rRr)r-s"r"r.rsyr0c rR.pVPRVP44VPR4V#)zdef load_args(reader):c3,"TF pRV 2xK R#5i)rNr)r ls& r"r#$InputWriter.lines..s1[4s[szload_args._version = 0)rrr)r@rs& r"linesInputWriter.liness8 $  1T[[11 )*r0 device_hintNr c(V^8dQhRRRRRRRR/#) r(rMrrrr rr+r*r)r-s"r"r.rs2$$'$? $ ' $ $r0c  \V4pVPPV4pVeV#R\VP4 2pRp\ R4\ V48wdRV: 2pRpVP pVPR8XdVf \R4hTp\R4V8wdRV: 2pVP4p \V 4p Rp VPe7VP PR8wdVPPV4p VPPV RV : RV V V R 24WPPV&V#) Nbufrz , dtype_hint=metaz:device_hint must be provided when storage device is 'meta'z , device=z = reader.storage(rr)r rrrrrIrrrrHrrrB write_storagerr) r@rMrr wsvmaybe_dtype_hintrrr nbytes_sourcers &&$$ r"rInputWriter.storagesXO ,    " "2 & =H$t++,- . T "&7 &C C!.zn=   '' ;;& "$P!F d #v -&vj1L '')1&9  :: !o&<&<&A&AV&K::33ODL c#L#32m_\NScRdde f "#2r0c$V^8dQhRRRRRR/#)r(r\r*rcrSr+r,r)r-s"r"r.rs!  3 < D r0c  ^RIHpHpVPVP 4VP VP R7p.pV!V!\RVPR7VP444'g2VP\\VP4444\R4VP 8wd VPRVP : 24V!\R4VP48H4'g$VPRVP4: 24\ P"P%V4pV'd'VP'RVP)444\+R4VP,8wd VPRVP,: 24\ P.P0P3V4p\5R4V8wdVPR V: 24VP6PR R P9V\\VP44.VO4,R V 2,4R#) r)statically_known_truesym_eq)r rNrFrzstorage_offset=c36"TFwrV RV: 2xK R#5i)=Nr)r krs& r"r#%InputWriter.tensor..sI1H1#Qqe 1Hszrequires_grad=zis_leaf=zreader.tensor(r) # )rrrrrMrrrr rrr*tuplerIrVrOr~raget_tensor_metadatarrrXrr^r_r`rWrrG) r@r\rcrrrrtensor_metadatarPs &&& r"rInputWriter.tensorsW,,   AGG $ %d!'':AHHJ G   KKE!((*-. / T "agg - KK& , -$ &t ,0@0@0B B   KK/!*:*:*<)?@ A,,::1=  KKI1F1F1HI I $T *aoo = KK.(;< =##..;;A> t $ / KK(7+. /  ii#eAGGn"5==> ?dVn  r0c$V^8dQhRRRRRR/#)r(r\r*argrr+r,r)r-s"r"r.r s!&&&#&$&r0c 6VPPRV: R\V4 24\V\\ 34'dVPPR4\ V4Fwr4V RV R2p\V\P4'dVPWT4K@\V\\P34'dVPWT4KyVPWT4K VPPR4R#R#)zreader.unsupported(z#) # unsupported type for dumping: z"""[]N)rrrr}rr enumerater~r rrr~r!r+)r@r\rrKaname_is&&& r"r+InputWriter.unsupported s !$)LTRUYK X cD%= ) ) KK  u %!# 61#Qa..KK*C#677KK*$$V/' KK  u % *r0c V^8dQhRRRR/#r$r)r-s"r"r.r4s  # $ r0c LVPPRV: RV R24R#)z reader.const(rz!, filtered out during compilationNrrr&s&&r"r'InputWriter.const4s' D85.O P r0c$V^8dQhRRRRRR/#)r(r\r*rgrr+r,r)r-s"r"r.r:s&DD3DSDTDr0c D\V\P4'd^\VPP 4pVPP pVPPRV: RV: RV 24R#VPPRV: RV 24R#)zreader.symint(z, expr=rN) r}r~r~r*rrehintrr)r@r\rgexpr_strrs&&& r"r!InputWriter.symint:sw c5<< ( (388==)H88==D KK  xwxl%PTvV W KK  weD6B Cr0c$V^8dQhRRRRRR/#)r(r\r*rztorch._C.Generatorr+r,r)r-s"r"r.rBs"  c (: t r0c VPpVPPRVP: RVP: RV 24R#)zreader.generator(rrN)rrrrindex)r@r\rrs&&& r"r1InputWriter.generatorBs<  b0@dV L r0c$V^8dQhRRRRRR/#)r(r\r*r5r+r,r)r-s"r"r.rHs&NN3N3N4Nr0c JVPPRV: RV 24R#)zreader.opaque(rNr)r@r\r5s&&&r"r8InputWriter.opaqueHs# ^,=+@dVLMr0)rr?rrrB)r`rarbrcrBrrrr+r'r!r1r8rdrr0r"rrsV ;E ;$BF $ *. $L @&( D NNr0rc ,V^8dQhRRRRRRRRR R /#) r(funcz&Callable[[list[Tensor]], list[Tensor]]rr* sym_shapeszdict[str, int] | Nonedefault_sym_shaperr+zdict[str, Any]r)r-s"r"r.r.LsAdd 0d d&d" d  dr0c aaaa^RIHpVP4UUu/uFwrVWebK pppRPVP 44p\ P !V4p RV R2p RV R2p Rp !RR 4p /pT;'g/oR VV3R lloR VV3R llpVPpVP!4FwppVR8XdK\P!V V4pV'dCVP4wpp\VPR44pVV,pV!VV4VV&\P!V V4pV'gKS!VP^44VV&K R\ P!V4P9dvV !4pVVR&\P !W4FOpVP4wpppp\VPR44pVV,p\#VVV!VV44KQ V#uuppi)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) ) dtype_abbrs|z(_tensor_constant\d+): \"(z0)\[\s*(.*?)\s*\]\" = self\.(_tensor_constant\d+)rz)\[\s*(.*?)\s*\]z Sym\((s\d+)\)c]tRtRtRtRtR#)/aot_graph_input_parser..TensorContaineripz#Container for tensors as attributesrN)r`rarbrc__doc__rdrr0r"TensorContainerrps-r0rc V^8dQhRRRR/#)r(r!r*r+rr)r-s"r"r.,aot_graph_input_parser..__annotate__xs>>C>C>r0c~<a\P!SS9;'gSRJV3Rl4SPSS4#)Nc<S R2#)z; not in symbolic_shapes and default sym shape not passed inr)r!sr"r=aot_graph_input_parser..get_sym_int..{s vhYZr0)r~_checkr)r!rsym_shapes_dictsfr" get_sym_int+aot_graph_input_parser..get_sym_intxs> o % F F):$)F Z ""6+<==r0c$V^8dQhRRRRRR/#)r(r rrrr+r r)r-s"r"r.rs#7 PVr0c<.p.p\V4FlwrEVP4pRV9d-S !V4pVPV4VPV4KHV'gKRVP\V44Kn VP'd\ P M\ PpV!W!S R7pVF"p \ PPW4K$ V#)rY)rr) rstriprrrr~randnzerosr mark_dynamic) r rresolved_shape dynamic_dimsrKdimrY constructorrrrrs && r" gen_tensor*aot_graph_input_parser..gen_tensors &FA))+Ccz$%%a(##A&3"))#c(3'&+%<%<%>*&&K(..0z H   , 3 #(<<> Iy)//#./Ei(E&ue4F5M /:6 5' A7F5M1"""4(333#% "v[[!8AE16 .Iy)Q)//#./Ei(E Iy*UE*B C B M[)s Hc V^8dQhRRRR/#)r(r)r*r+zCallable[[T], T]r)r-s"r"r.r.s##c#&6#r0caa\P!4o\PP \PP S44oRV3RllpRVV3Rllp\ P!V4V#)z Decorator to cProfile a given function and save the result to disk on process exit. Args: filename: filename to save profile to c V^8dQhRRRR/#)r(fnrr+r)r-s"r"r.%profile_to_file..__annotate__s  c c r0cP<a\P!S4RVV3Rll4pV#)c$V^8dQhRRRRRR/#)r(rrrr+r)r-s"r"r.8profile_to_file..decorator..__annotate__s!  3 # # r0c<SP4S!V/VBSP4# SP4i;ir)enabledisable)rrrprofs*,r"wrapper3profile_to_file..decorator..wrappers1 KKM 4*6*  s+=) functoolswraps)rr r sf r" decorator"profile_to_file..decorators*      r0cV^8dQhRR/#rr)r-s"r"r.rs   T  r0c <SPS4\PP\P !RS RS R244R#)z! Wrote profile to z+, view with: snakeviz z N) dump_statssysstderrrXrIrJ)r)r sr"save_it profile_to_file..save_itsK !  OO""*,&Z(  r0)cProfileProfiler6r7r9 expanduseratexitregister)r)rrr sf @r"profile_to_filers\    Dwwrww11(;r<s$#   #44 ##.@.;3M992,! CL 89  $ $ &     J $$..==?GG SWXJ yy C CC  012 644n ~ eWeWP  6 e @ED@ V I  0@7  7 77t0(1    RV/"%--0"5<<#67*1-)%0#E*FeehKNKN\dN#r0