Method getEmbeddingModel

embedding/embeddingHelper.py:481–523

model_type: [ default, all-MiniLM-L6-v2, Sentence Transformers, OpenAI, Cohere, Instructor models, Google PaLM API models, HuggingFace ]

(self, model_type, OPENAI_API_KEY=None, OPENAI_API_BASE_PATH=None, Cohere_API_KEY=None,
                          Sentence_Transformersmodel_name=None, Cohere_model_name=None, Instructor_model_name=None,
                          Google_api_key=None, Google_model_name=None, HuggingFace_api_key=None,
                          HuggingFace_model_name=None)

        return self.embedding_model([text])

    def getEmbeddingModel(self, model_type, OPENAI_API_KEY=None, OPENAI_API_BASE_PATH=None, Cohere_API_KEY=None,
                          Sentence_Transformersmodel_name=None, Cohere_model_name=None, Instructor_model_name=None,
                          Google_api_key=None, Google_model_name=None, HuggingFace_api_key=None,
                          HuggingFace_model_name=None):
        """
        model_type:
        [ default,
        all-MiniLM-L6-v2,
        Sentence Transformers,
        OpenAI,
        Cohere,
        Instructor models,
        Google PaLM API models,
        HuggingFace ]
        """
        logging.info(self.embedding_type)
        model = embedding_functions.DefaultEmbeddingFunction()
        if model_type == "all-MiniLM-L6-v2":
            model = embedding_functions.DefaultEmbeddingFunction()
        elif model_type == "Sentence Transformers":
            model = embedding_functions.SentenceTransformerEmbeddingFunction(model_name=Sentence_Transformersmodel_name)
        elif model_type == "OpenAI":
            model = embedding_functions.OpenAIEmbeddingFunction(
                api_key=OPENAI_API_KEY,
                api_base=OPENAI_API_BASE_PATH,
                model_name="text-embedding-ada-002"
            )
        elif model_type == "Cohere":
            model = embedding_functions.CohereEmbeddingFunction(api_key=Cohere_API_KEY, model_name=Cohere_model_name)
        elif model_type == "Instructor models":
            if torch.cuda.is_available():
                model = embedding_functions.InstructorEmbeddingFunction(
                    model_name=Instructor_model_name, device="cuda")
            else:
                model = embedding_functions.InstructorEmbeddingFunction()
        elif model_type == "Google PaLM API models":
            model = embedding_functions.GooglePalmEmbeddingFunction(api_key=Google_api_key)
        elif model_type == "HuggingFace":
            model = embedding_functions.HuggingFaceEmbeddingFunction(
                api_key=HuggingFace_api_key,
                model_name=HuggingFace_model_name
            )
        return model

    def insert_cahing(self, query: list = None, query_vec: list = None, result=None):
        # default_ef = embedding_functions.DefaultEmbeddingFunction()