Skip to content

Embedding Models API

EmbeddingModel

Bases: SynalinksSaveable

An embedding model API wrapper.

Embedding models are a type of machine learning model used to convert high-dimensional data, such as text into lower-dimensional vector representations while preserving the semantic meaning and relationships within the data. These vector representations, known as embeddings, allow for more efficient and effective processing in various tasks.

Many providers are available like OpenAI, Azure OpenAI, Vertex AI or Ollama.

For the complete list of models, please refer to the providers documentation.

Using OpenAI models

import synalinks
import os

os.environ["OPENAI_API_KEY"] = "your-api-key"

embedding_model = synalinks.EmbeddingModel(
    model="openai/text-embedding-ada-002",
)

Using Azure OpenAI models

import synalinks
import os

os.environ["AZURE_API_KEY"] = "your-api-key"
os.environ["AZURE_API_BASE"] = "your-api-base"
os.environ["AZURE_API_VERSION"] = "your-api-version"

embedding_model = synalinks.EmbeddingModel(
    model="azure/<your_deployment_name>",
)

Using VertexAI models

import synalinks
import os

embedding_model = synalinks.EmbeddingModel(
    model="vertex_ai/text-embedding-004",
    vertex_project = "hardy-device-38811", # Your Project ID
    vertex_location = "us-central1",  # Project location
)

Using Ollama models

import synalinks

embedding_model = synalinks.EmbeddingModel(
    model="ollama/mxbai-embed-large",
)

Parameters:

Name Type Description Default
model str

The model to use.

 None
api_base str

Optional. The endpoint to use.

 None
retry int

Optional. The number of retry.

 5
Source code in synalinks/src/embedding_models/embedding_model.py 
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
@synalinks_export(
    ["synalinks.EmbeddingModel", "synalinks.embedding_models.EmbeddingModel"]
)
class EmbeddingModel(SynalinksSaveable):
    """An embedding model API wrapper.

    Embedding models are a type of machine learning model used to convert
    high-dimensional data, such as text into lower-dimensional vector
    representations while preserving the semantic meaning and relationships
    within the data. These vector representations, known as embeddings,
    allow for more efficient and effective processing in various tasks.

    Many providers are available like OpenAI, Azure OpenAI, Vertex AI or Ollama.

    For the complete list of models, please refer to the providers documentation.

    **Using OpenAI models**

    ```python
    import synalinks
    import os

    os.environ["OPENAI_API_KEY"] = "your-api-key"

    embedding_model = synalinks.EmbeddingModel(
        model="openai/text-embedding-ada-002",
    )
    ```

    **Using Azure OpenAI models**

    ```python
    import synalinks
    import os

    os.environ["AZURE_API_KEY"] = "your-api-key"
    os.environ["AZURE_API_BASE"] = "your-api-base"
    os.environ["AZURE_API_VERSION"] = "your-api-version"

    embedding_model = synalinks.EmbeddingModel(
        model="azure/<your_deployment_name>",
    )
    ```

    **Using VertexAI models**

    ```python
    import synalinks
    import os

    embedding_model = synalinks.EmbeddingModel(
        model="vertex_ai/text-embedding-004",
        vertex_project = "hardy-device-38811", # Your Project ID
        vertex_location = "us-central1",  # Project location
    )
    ```

    **Using Ollama models**

    ```python
    import synalinks

    embedding_model = synalinks.EmbeddingModel(
        model="ollama/mxbai-embed-large",
    )
    ```

    Args:
        model (str): The model to use.
        api_base (str): Optional. The endpoint to use.
        retry (int): Optional. The number of retry.
    """

    def __init__(self, model=None, api_base=None, retry=5):
        if model is None:
            raise ValueError(
                "You need to set the `model` argument for any EmbeddingModel"
            )
        self.model = model
        if self.model.startswith("ollama") and not api_base:
            self.api_base = "http://localhost:11434"
        else:
            self.api_base = api_base
        self.retry = retry

    async def __call__(self, texts, **kwargs):
        """
        Call method to get dense embeddings vectors

        Args:
            texts (list): A list of texts to embed.

        Returns:
            (list): The list of corresponding vectors.
        """
        for i in range(self.retry):
            try:
                if self.api_base:
                    response = litellm.embedding(
                        model=self.model,
                        input=texts,
                        api_base=self.api_base,
                        **kwargs,
                    )
                else:
                    response = litellm.embedding(
                        model=self.model,
                        input=texts,
                        **kwargs,
                    )
                vectors = []
                for data in response["data"]:
                    vectors.append(data["embedding"])
                return {"embeddings": vectors}
            except Exception as e:
                warnings.warn(f"Error occured while trying to call {self}: " + str(e))
        raise RuntimeError(
            f"Failed to retrieve embeddings with {self} after {self.retry} attempts."
        )

    def _obj_type(self):
        return "EmbeddingModel"

    def get_config(self):
        return {
            "model": self.model,
            "api_base": self.api_base,
            "retry": self.retry,
        }

    @classmethod
    def from_config(cls, config):
        return cls(**config)

    def __repr__(self):
        api_base = f" api_base={self.api_base}" if self.api_base else ""
        return f"<EmbeddingModel model={self.model}{api_base}>"

__call__(texts, **kwargs) async

Call method to get dense embeddings vectors

Parameters:

Name Type Description Default
texts list

A list of texts to embed.

 required

Returns:

Type Description
list

The list of corresponding vectors.
Source code in synalinks/src/embedding_models/embedding_model.py 
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
async def __call__(self, texts, **kwargs):
    """
    Call method to get dense embeddings vectors

    Args:
        texts (list): A list of texts to embed.

    Returns:
        (list): The list of corresponding vectors.
    """
    for i in range(self.retry):
        try:
            if self.api_base:
                response = litellm.embedding(
                    model=self.model,
                    input=texts,
                    api_base=self.api_base,
                    **kwargs,
                )
            else:
                response = litellm.embedding(
                    model=self.model,
                    input=texts,
                    **kwargs,
                )
            vectors = []
            for data in response["data"]:
                vectors.append(data["embedding"])
            return {"embeddings": vectors}
        except Exception as e:
            warnings.warn(f"Error occured while trying to call {self}: " + str(e))
    raise RuntimeError(
        f"Failed to retrieve embeddings with {self} after {self.retry} attempts."
    )