embantic


Nameembantic JSON
Version 0.1.0 PyPI version JSON
download
home_pageNone
SummaryEmbed pydantic models
upload_time2024-10-02 20:31:14
maintainerNone
docs_urlNone
authorPatrick Barker
requires_python<4.0,>=3.10
licenseMIT
keywords
VCS
bugtrack_url
requirements No requirements were recorded.
Travis-CI No Travis.
coveralls test coverage No coveralls. 
            # emdantic

Embed pydantic models

Embantic is an OVM (Object Vector Mapper) which embeds pydantic models to vectors allowing for semantic search and retrieval of arbitrary mutlimodal objects.

## Installation

```sh
pip install embantic
```

## Usage

Create a data model you would like to embed

```python
from emdantic import EmbModel

class Foo(EmbModel):
    a: int
    b: str
    c: float
    d: bool
    e: list[int]
    f: dict[str, int]
    g: Image.Image
    h: Optional[EmbModel]
```

Embed the data model

```python
foo = Foo(
    a=1, 
    b="hello", 
    c=3.14, 
    d=True, 
    e=[1, 2, 3], 
    f={"a": 1, "b": 2},
    g=Image.open("path/to/image.png")
)

vectors = foo.embed()
```

Or embed the data model and store the embeddings in a vector database

```python
foo.store()
```

Search for objects by text

```python
results: List[Foo] = Foo.search("hello")
```

Search for objects by image

```python
results: List[Foo] = Foo.search(Image.open("path/to/image.png"))
```

Specify the embedding models to use

```python
class Foo(EmbModel):
    __text_model__ = "text-embedding-ada-002"
    __image_model__ = "SigLIP-400M"

    ...
```

## Backends

### Vector Databases

Vector database backends are configured using `EMDANTIC_VECTOR_BACKEND` environment variable. Supported backeends are currently: `faiss`


### Embedding Models

Embedding models are configured using the `__text_model__` and `__image_model__` parameters on the `EmbModel` class. Supported models are currently: `text-embedding-ada-002` and `SigLIP-400M`.

Alternatively, you can set the `EMDANTIC_TEXT_EMBEDDING_MODEL` and `EMDANTIC_IMAGE_EMBEDDING_MODEL` environment variable as the model defaults.

### Database

Embdantic also uses a database to store the raw objects in JSON format. The database is configured using the `EMDANTIC_DATABASE_BACKEND` environment variable. Supported databases are currently: `sqlite`.

Raw data

            {
    "_id": null,
    "home_page": null,
    "name": "embantic",
    "maintainer": null,
    "docs_url": null,
    "requires_python": "<4.0,>=3.10",
    "maintainer_email": null,
    "keywords": null,
    "author": "Patrick Barker",
    "author_email": "patrickbarkerco@gmail.com",
    "download_url": "https://files.pythonhosted.org/packages/a2/ce/c60abb07bc34cf677b26140d34439b0e435512131d49d3126e588a53ae94/embantic-0.1.0.tar.gz",
    "platform": null,
    "description": "# emdantic\n\nEmbed pydantic models\n\nEmbantic is an OVM (Object Vector Mapper) which embeds pydantic models to vectors allowing for semantic search and retrieval of arbitrary mutlimodal objects.\n\n## Installation\n\n```sh\npip install embantic\n```\n\n## Usage\n\nCreate a data model you would like to embed\n\n```python\nfrom emdantic import EmbModel\n\nclass Foo(EmbModel):\n    a: int\n    b: str\n    c: float\n    d: bool\n    e: list[int]\n    f: dict[str, int]\n    g: Image.Image\n    h: Optional[EmbModel]\n```\n\nEmbed the data model\n\n```python\nfoo = Foo(\n    a=1, \n    b=\"hello\", \n    c=3.14, \n    d=True, \n    e=[1, 2, 3], \n    f={\"a\": 1, \"b\": 2},\n    g=Image.open(\"path/to/image.png\")\n)\n\nvectors = foo.embed()\n```\n\nOr embed the data model and store the embeddings in a vector database\n\n```python\nfoo.store()\n```\n\nSearch for objects by text\n\n```python\nresults: List[Foo] = Foo.search(\"hello\")\n```\n\nSearch for objects by image\n\n```python\nresults: List[Foo] = Foo.search(Image.open(\"path/to/image.png\"))\n```\n\nSpecify the embedding models to use\n\n```python\nclass Foo(EmbModel):\n    __text_model__ = \"text-embedding-ada-002\"\n    __image_model__ = \"SigLIP-400M\"\n\n    ...\n```\n\n## Backends\n\n### Vector Databases\n\nVector database backends are configured using `EMDANTIC_VECTOR_BACKEND` environment variable. Supported backeends are currently: `faiss`\n\n\n### Embedding Models\n\nEmbedding models are configured using the `__text_model__` and `__image_model__` parameters on the `EmbModel` class. Supported models are currently: `text-embedding-ada-002` and `SigLIP-400M`.\n\nAlternatively, you can set the `EMDANTIC_TEXT_EMBEDDING_MODEL` and `EMDANTIC_IMAGE_EMBEDDING_MODEL` environment variable as the model defaults.\n\n### Database\n\nEmbdantic also uses a database to store the raw objects in JSON format. The database is configured using the `EMDANTIC_DATABASE_BACKEND` environment variable. Supported databases are currently: `sqlite`.",
    "bugtrack_url": null,
    "license": "MIT",
    "summary": "Embed pydantic models",
    "version": "0.1.0",
    "project_urls": null,
    "split_keywords": [],
    "urls": [
        {
            "comment_text": "",
            "digests": {
                "blake2b_256": "084c3cc0fed7b0fa84862c971b2dd65062a6fa654ddfafd0c06c97fa66c1319a",
                "md5": "8d0aad905057dfc69cc10716ada6a044",
                "sha256": "34b6ec6352c24349b948ef1d5e51a3e9c824bc654653fb3c9b791854df253056"
            },
            "downloads": -1,
            "filename": "embantic-0.1.0-py3-none-any.whl",
            "has_sig": false,
            "md5_digest": "8d0aad905057dfc69cc10716ada6a044",
            "packagetype": "bdist_wheel",
            "python_version": "py3",
            "requires_python": "<4.0,>=3.10",
            "size": 8023,
            "upload_time": "2024-10-02T20:31:12",
            "upload_time_iso_8601": "2024-10-02T20:31:12.854722Z",
            "url": "https://files.pythonhosted.org/packages/08/4c/3cc0fed7b0fa84862c971b2dd65062a6fa654ddfafd0c06c97fa66c1319a/embantic-0.1.0-py3-none-any.whl",
            "yanked": false,
            "yanked_reason": null
        },
        {
            "comment_text": "",
            "digests": {
                "blake2b_256": "a2cec60abb07bc34cf677b26140d34439b0e435512131d49d3126e588a53ae94",
                "md5": "9728ab223e3baf6c5b3f34460f30e8f1",
                "sha256": "bff517675bc439349ffde5229bf71d771ad06d58e255be17cccb8ce41025ec2f"
            },
            "downloads": -1,
            "filename": "embantic-0.1.0.tar.gz",
            "has_sig": false,
            "md5_digest": "9728ab223e3baf6c5b3f34460f30e8f1",
            "packagetype": "sdist",
            "python_version": "source",
            "requires_python": "<4.0,>=3.10",
            "size": 7133,
            "upload_time": "2024-10-02T20:31:14",
            "upload_time_iso_8601": "2024-10-02T20:31:14.427613Z",
            "url": "https://files.pythonhosted.org/packages/a2/ce/c60abb07bc34cf677b26140d34439b0e435512131d49d3126e588a53ae94/embantic-0.1.0.tar.gz",
            "yanked": false,
            "yanked_reason": null
        }
    ],
    "upload_time": "2024-10-02 20:31:14",
    "github": false,
    "gitlab": false,
    "bitbucket": false,
    "codeberg": false,
    "lcname": "embantic"
}
Patrick Barker
Elapsed time: 0.33382s