>:j#SSKJr SSKrSSKrSSKJr SSKJr SSKJ r J r J r \R"\ 5rSSjrSSjrg) ) annotationsN)Path)PretrainedConfig)_save_pretrained_wrapperbackend_should_exportbackend_warn_to_savec @SSKnSSKJnJnJnJnJn UUU US.n X*;a0SRU R55n [SUSU 35eXn URS UR5S5US '[U5n U R5nS nS n[!XX5UU5unnU(aURS S5 U R""U4UUS .UD6n[%UR&SS9UlU(a [)XU5 U$![a [S5ef=f)a Load and perhaps export an ONNX model using the Optimum library. Args: model_name_or_path (str): The model name on Hugging Face (e.g. 'naver/splade-cocondenser-ensembledistil') or the path to a local model directory. config (PretrainedConfig): The model configuration. task_name (str): The task name for the model (e.g. 'feature-extraction', 'fill-mask', 'sequence-classification'). model_kwargs (dict): Additional keyword arguments for the model loading. rN)ONNX_WEIGHTS_NAMEORTModelForCausalLMORTModelForFeatureExtractionORTModelForMaskedLM!ORTModelForSequenceClassificationzfeature-extractionz fill-maskzsequence-classificationztext-generation, Unsupported task: . Supported tasks: zUsing the ONNX backend requires installing Optimum and ONNX Runtime. You can install them with pip: `pip install sentence-transformers[onnx]` or `pip install sentence-transformers[onnx-gpu]`providerONNXz*.onnx file_nameconfigexportonnx subfolder) onnxruntimeoptimum.onnxruntimer r r r rjoinkeys ValueErrorModuleNotFoundError Exceptionpopget_available_providersrexistsrfrom_pretrainedr_save_pretrainedr)model_name_or_pathr task_name model_kwargsortr r r r rtask_to_model_mappingsupported_tasks model_cls load_pathis_local backend_nametarget_file_globrmodels _/home/mande/repo/quber/.venv/lib/python3.13/site-packages/sentence_transformers/backend/load.pyload_onnx_modelr5sf !  #?,'H2 !   1"ii(=(B(B(DEO1)NP\FL d+  % %     E6e6L6LX^_E/<H LQ   ?   s ADDc SSKJnJnJnJnJn UUUUS.n X);a0SR U R55n [SUSU 35eXn [U5n U R5n SnS n[XX4X5unnU(aURS S 5 S U;aqUS n[U[ 5(dV[U5R5(d [S 5e[#USS9n[$R&"U5US 'S S S 5 OO0US 'U R("U4UUS.UD6n[+UR,SS9UlU(a [/X U5 U$![a [S5ef=f!,(df  Nk=f)a Load and perhaps export an OpenVINO model using the Optimum library. Args: model_name_or_path (str): The model name on Hugging Face (e.g. 'naver/splade-cocondenser-ensembledistil') or the path to a local model directory. config (PretrainedConfig): The model configuration. task_name (str): The task name for the model (e.g. 'feature-extraction', 'fill-mask', 'sequence-classification'). model_kwargs (dict): Additional keyword arguments for the model loading. r)OV_XML_FILE_NAMEOVModelForCausalLMOVModelForFeatureExtractionOVModelForMaskedLM OVModelForSequenceClassificationrrrrzUsing the OpenVINO backend requires installing Optimum and OpenVINO. You can install them with pip: `pip install sentence-transformers[openvino]`OpenVINOz openvino*.xmlrN ov_configzXov_config should be a dictionary or a path to a .json file containing an OpenVINO configzutf-8)encodingropenvinor)optimum.intel.openvinor7r8r9r:r;rrr r!r"rr%rr# isinstancedictopenjsonloadr&rr'r)r(rr)r*r7r8r9r:r;r,r-r.r/r0r1r2rr=fr3s r4load_openvino_modelrG\s   #>+'G1 !   1"ii(=(B(B(DEO1)rSs7" =uu   8 $K\SrR