+ G-j%^RIt^RIt^RIt^RIHt^RIt^RIt^RIHt^RI H t H t ^RI H t ^RIHt^RIHtHt]P&!]4tR RR lltR#) N)Path)SymbolicShapeInference)extract_raw_data_from_modelhas_external_data)ReplaceUpsampleWithResize) ONNXModel)add_pre_process_metadata&save_and_reload_model_with_shape_inferc8V^8dQhR\\,\P,R,R\\,R,R\R\R\R\R\ R \R \ R \R \R \R,R\ RR/#) input_modelNoutput_model_pathskip_optimizationskip_onnx_shapeskip_symbolic_shape auto_mergeint_maxguess_output_rankverbosesave_as_external_dataall_tensors_to_one_fileexternal_data_locationexternal_data_size_thresholdreturn)strronnx ModelProtoboolint)formats"y/Volumes/fast/ai/experiments/nudenet-smoke/.venv/lib/python3.14/site-packages/onnxruntime/quantization/shape_inference.py __annotate__r"sr,r,tdoo-4r,TzD(r,r, r,  r,  r,r,r,r, r,"r, $Jr,#&r, r,c  VfV PRR4pVfQhVfQR4h\P!RR7;_uu_4p\V4p\ V\ P 4'dTM\ P!V4pVPUu.uF*pVP'dVPR8XgK(VNK, pp\V4^8XdiV^,PpV^ 8:dO\\V4V4P4\ PP!V^ 4p\#V4pV'g0\$P'R4\(P*!VVVVV4pV'EgeV'gQ\-VR, 4pV 'd\ P.!VVR V V R R 7M\ P0!VV4Rp\-VR , 4p\2P4!4pVVn\2P8P:Vn\ V\ P 4'd`\?V4'd \AR 4h\CV4wppVPE\GV4\GV44VPI4pM"V'dV 'dVPKRR4\2PL!VVR.R7p?TpV'gVeQ\-VR, 4pV 'd\ P.!VVR V V R R 7M\ P0!VV4Rp\ V\ P 4'd8\-\V4R, 4p\ P.!VVR V V R R 7\-VR, 4p\ PVPYVV4\ P!V4pRRR4Xf9\ V\ P 4'dTM\ P!V4p\[V4V 'd \ P.!VVR V V V R R7R#\ P0!VV4R#uupi \NdB\$PQR4\$PQ\RPT!44ELi;i +'giL;i)a(Shape inference and model optimization, in preparation for quantization. Args: input_model: Path to the input model file or ModelProto output_model_path: Path to the output model file skip_optimization: Skip model optimization step if true. This may result in ONNX shape inference failure for some models. skip_onnx_shape: Skip ONNX shape inference. Symbolic shape inference is most effective with transformer based models. Skipping all shape inferences may reduce the effectiveness of quantization, as a tensor with unknown shape can not be quantized. skip_symbolic_shape: Skip symbolic shape inference. Symbolic shape inference is most effective with transformer based models. Skipping all shape inferences may reduce the effectiveness of quantization, as a tensor with unknown shape can not be quantized. auto_merge: For symbolic shape inference, automatically merge symbolic dims when conflict happens. int_max: For symbolic shape inference, specify the maximum value for integer to be treated as boundless for ops like slice guess_output_rank: Guess output rank to be the same as input 0 for unknown ops verbose: Logs detailed info of inference, 0: turn off, 1: warnings, 3: detailed save_as_external_data: Saving an ONNX model to external data all_tensors_to_one_file: Saving all the external data to one file external_data_location: The file location to save the external file external_data_size_threshold: The size threshold for external data Ninput_model_pathzoutput_model_path is required.z pre.quant.)prefixzai.onnxz&Performing symbolic shape inference...zsymbolic_shape_inferred.onnxTF)rrsize_thresholdconvert_attributezoptimized.onnxzModelProto has external data not loaded into memory, ORT cannot create session. Please load external data before calling this function. See https://onnx.ai/onnx/repo-docs/ExternalData.html for more information.z7session.optimized_model_external_initializers_file_namezoptimized.onnx.dataCPUExecutionProvider) providerszYONNX Runtime Model Optimization Failed! Consider rerun with option `--skip_optimization'.zmodel_input.onnxzonnx_shape_inferred.onnx)rrlocationr'r().poptempfileTemporaryDirectoryr isinstancerrload opset_importdomainlenversionrrapplyversion_converterconvert_versionr loggerinfor infer_shapesr save_modelsave onnxruntimeSessionOptionsoptimized_model_filepathGraphOptimizationLevelORT_ENABLE_BASICgraph_optimization_levelr ValueErrorradd_external_initializerslistSerializeToStringadd_session_config_entryInferenceSession Exceptionerror traceback format_excshape_inferenceinfer_shapes_pathr )r rrrrrrrrrrrrdeprecated_kwargs quant_tmp_dir temp_pathmodelopsetai_onnx_domain opset_versionopt_model_path sess_optionexternal_namesexternal_valuessessinferred_model_paths&&&&&&&&&&&&&, r!quant_pre_processr\sV'++,>E  "" "  (J*JJ (  $ $L 9 9]' )+tGG TYYWbMc .3-?-?q-?Eu|||W\WcWcgpWp%%-?q ~ ! #*1-55M"))E*:MJPPR..>>ubI>uE" KK@ A*77! E! &!).L"LM (OO#.20G'C*/ IIe[1 -=!=>N 5)88: 7E 47B7Y7Y7j7j 4k4??;;(55(i 7RR]6^3NO99$~:NPTUdPef"-"?"?"AK)-B88QSh#33KYoXpq)K  !).L"LM (OO#.20G'C*/ IIe[1+t77!$}"58J"JK *.,C#?&+ #&i2L&L"M   2 2;@S TII12E_ :b })+tGG TYYWbMcU#   "&$;+7#  %*+ur~ 5 o Y1134  5M : 9sxASQ10Q1Q1 BS 7SS A"S.CQ60Q68-Q6% S/C3S1S6AS>SSS S ) NNFFFFiFrFFNi)loggingr-rKpathlibrrr=&onnxruntime.tools.symbolic_shape_inferr#onnxruntime.transformers.onnx_utilsrrfusionsr onnx_modelr quant_utilsr r getLogger__name__r8r\r#r!rgsD I^.!Y   8 $r,r,r#