hbSSKrSSKJr SSKrSrSrS\R R4Sjrg)N)abccj^ ^ U U 4Sjm /m SnSn[U5GHupVUR[RR:XaC[ U[ U55H(nURSU3nUS- nXn UT "X5- nM* MqUR[RR:XdPUR[RR:Xd(UR[RR:XaVURnSn XS- nSn U[ U5:a XbXn OX;a X(bX(n Sn T "X5n U (aXK- nGM<GM?UR[RR:XdGMjUR5HupUT ;dM T "X5 M GM T $)NcV>UcgSn[U[R5(a'[U5Hup4UT"USU3U5- nM U$[U[R5(a,UR 5HupTUT"USU3U5- nM U$TR U5 g)z;Returns number of expanded inputs that _add_input processedr_) isinstancerSequence enumerateMappingitemsappend)nameinputnum_expanded_non_none_inputsivalkey _add_input input_namess b/var/www/fran/franai/venv/lib/python3.13/site-packages/onnxruntime/tools/pytorch_export_helpers.pyr1_parse_inputs_for_onnx_export.._add_input s ='($ eS\\ * *$E*- dV1QC=#0NN,+0 / s{{ + +"KKM, dV1SE?C0PP,* 0 / 4 rrrTF) r kindinspect ParameterVAR_POSITIONALrangelenrPOSITIONAL_ONLYPOSITIONAL_OR_KEYWORD KEYWORD_ONLY VAR_KEYWORDr )all_input_parametersinputskwargsvar_positional_idx'num_expanded_non_none_positional_inputs input_idxinput_parameterargs_irinp is_positional"num_expanded_non_none_inputs_localrrs @@r_parse_inputs_for_onnx_exportr. s"HK./+&/0D&E"   7#4#4#C#C C 3v;7)../q1C0DE"a'"n7:d;PP7 8  G$5$5$E$E E##w'8'8'N'NN##w'8'8'E'EE#''DC  +I M3v;&6+<+H'FL$<l % 1;D1F .7]7  ! !W%6%6%B%B B#\\^ {*t),9'F@ rc$SnSnUVs/sHoS"U5(aU"U5OUPM nnUUVs/sH'owU;dM U"X'5(a U"X'5OX'PM) sn- nU(dUR05 [U5$s snfs snf)z5Flatten args and kwargs in a single tuple of tensors.c<[U5[[[1;$N)typeintboolfloatvalues ris_primitive_type0_flatten_module_input..is_primitive_type\sE{sD%000rc.[R"U5$r1)torchtensorr6s r to_tensor(_flatten_module_input..to_tensor_s||E""r)r tuple)namesargsr%r8r=argretrs r_flatten_module_inputrDXs1#IM M.s339S> <C Maf afY]rxjxT#4V\#B#B &, Taf C  2 : N s B B !B modulec[R"UR5RR 5n[ X1U5n[ XAU5nXE4$)a Infer the input names and order from the arguments used to execute a PyTorch module for usage exporting the model via torch.onnx.export. Assumes model is on CPU. Use `module.to(torch.device('cpu'))` if it isn't. Example usage: input_names, inputs_as_tuple = infer_input_info(module, ...) torch.onnx.export(module, inputs_as_type, 'model.onnx', input_names=input_names, output_names=[...], ...) :param module: Module :param inputs: Positional inputs :param kwargs: Keyword argument inputs :return: Tuple of ordered input names and input values. These can be used directly with torch.onnx.export as the `input_names` and `inputs` arguments. )r signatureforward parametersvaluesr.rD)rEr$r%module_parametersrinputs_as_tuples rinfer_input_inforMosJ  ))&..9DDKKM/0A6RK+KHO  ''r) r collectionsrr;r.rDnnModulerMrrrRs0 K\.(UXX__(r