Skip to main content

AI Library to authenticate and manage LLM models

Project description

ai-factory-model

PyPI version Supported Python Versions Build Status PyPI - Downloads License

ai-factory-model is a modular Python library aimed at integrating with multiple language models (LLMs), cloud providers, and auxiliary utilities for development and infrastructure.

This library is designed to facilitate interaction with LLMs from OpenAI, Azure, Google, and Ollama, also integrating authentication, external configuration, Jinja2 templates, and reusable components.

Features

  • Support for multiple LLM models:
    • OpenAI (chat and embeddings)
    • Azure OpenAI
    • Google Generative AI
    • Ollama
    • LangChain and variants
  • Configuration modules (decouple, YAML)
  • Authentication via Azure Identity
  • Content generation via Jinja2 templates
  • Clear separation of responsibilities with modules such as:
    • logger
    • security
    • auth_clients
    • model_* (interfaces for different LLMs)

Installation

From PyPI:

pip install ai-factory-model

Setup

To use the model factory, you need to define a series of environment variables that allow connection to the various model hosting services:

AZURE_TENANT_ID = <id_tenant_azure>
AZURE_CLIENT_ID = <id_client_azure>
AZURE_CLIENT_SECRET = <secret_passphrase_azure_client>
AZURE_TOKEN_URL = <azure_url_token_generator>

For enhanced security, there is a connection to KeyVault. To define the connection to the corresponding key store, use:

KV_NAME = <kv_name>
KV_TENANT_ID = <id_kv_tenant>
KV_CLIENT_ID = <id_kv_client>
KV_SECRET = <secret_passphrase_kv>

With the KeyVault connection established, the values to be retrieved from the key store should be specified using the following nomenclature:

VARIABLE_SECRET = kv{name-of-secret-at-kv}

For example:

# Transition from having the secret in raw form
AZURE_CLIENT_SECRET = <secret_passphrase_azure_client>

# To retrieving it from the KV
AZURE_CLIENT_SECRET = kv{<name_secret_azure_client>}

Additionally, if you have a file containing the various model configurations you wish to use, you should specify it with the corresponding variable.

MODELS_CONFIG_FILE = <path_to_models_declarations_file>

Basic usage

Using prompt:

from ai_factory_model import ModelFactory

model = ModelFactory.get_model("azai_gtp4o")
params = ["Eres un guía turístico", "¿Dónde está Plasencia?"]

response = model.prompt(params=params)

print(type(response))
# Output:
# <class 'str'>

print(response)
# Output:
# Plasencia es una ciudad situada en la comunidad autónoma de Extremadura, en el oeste de España. Se encuentra en la provincia de Cáceres, a orillas del río Jerte. Plasencia está aproximadamente a unos 80 kilómetros al norte de la ciudad de Cáceres y a unos 250 kilómetros al oeste de Madrid. Es conocida por su casco histórico, que incluye la Catedral de Plasencia, y por su cercanía al Valle del Jerte, famoso por sus cerezos en flor.

Using langchain instance:

from ai_factory_model import ModelFactory

model = ModelFactory.get_model("azai_gtp4o")
params = ["Eres un guía turístico", "¿Cuál es la capital de España?"]

response = model.get_client.invoke([
    {"role": "system", "content": params[0]},
    {"role": "user", "content": params[1]}
])

print(type(response))
# Output:
# <class 'langchain_core.messages.ai.AIMessage'>

print(f"{response.content}")
# Output:
# La capital de España es Madrid. Es una ciudad vibrante y llena de historia, conocida por su rica cultura, su arquitectura impresionante y su animada vida nocturna. Además, Madrid alberga importantes museos como el Museo del Prado y el Museo Reina Sofía, así como el Palacio Real y el Parque del Retiro.

Render a template:

from ai_factory_model import ModelFactory, SEP_PATTERN

model = ModelFactory.get_model("azai_gtp4o")
params = {"system": "Eres un guía turístico", "user": "¿Qué visitar en Mérida de Extremadura?"}

template_content = (
    f"{{{{ system }}}}"
    f"{SEP_PATTERN}"
    f"{{{{ user }}}}"
)
prompt_template = Template(template_content)

response = model.prompt_render(
    template=prompt_template,
    params=params,
    sep_pattern=SEP_PATTERN
)

print(type(response))
# Output:
# <class 'str'>

print(f"{response}")
# Output:
# ¡Mérida es una ciudad fascinante llena de historia y patrimonio! Es conocida por su impresionante legado romano, ya que fue una de las ciudades más importantes de la antigua Hispania. Aquí tienes una lista de los lugares imprescindibles que deberías visitar en Mérida: [...]

Project structure

ai_factory_model/
├── ai_factory_model/
│   ├── __init__.py
│   ├── config/
│   ├── llm/
│   ├── logger/
│   ├── security/
│   ├── vectordb/
├── pyproject.toml
├── requirements.txt
└── README.md

Dependencies

This package requires the followind external libraries:

  • python-decouple
  • PyYAML
  • openai
  • jinja2
  • azure-core
  • azure-identity
  • azure-keyvault-secrets
  • langchain
  • langchain-openai
  • langchain-google-genai
  • langchain-community
  • langchain-azure-ai
  • langchain-ollama
  • langchain-cohere
  • azure-search-documents
  • psycopg[binary]

Requirements

  • Python 3.12 o superior
  • Credential access/API keys to your needed providers (OpenAI, Azure, etc.)

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

ai_factory_model-0.0.8.tar.gz (26.8 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

ai_factory_model-0.0.8-py3-none-any.whl (26.1 kB view details)

Uploaded Python 3

File details

Details for the file ai_factory_model-0.0.8.tar.gz.

File metadata

  • Download URL: ai_factory_model-0.0.8.tar.gz
  • Upload date:
  • Size: 26.8 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.13.2

File hashes

Hashes for ai_factory_model-0.0.8.tar.gz
Algorithm Hash digest
SHA256 785b737284b33c9521bbfebb2c4293ff20cd8dc75f86f1d7b9413e4088977d7d
MD5 015f625218bd37b794a4b7fffbd27147
BLAKE2b-256 a7104072b3b32f716d5cbf173068a86bfc7ab048025f5f65ace8fc1c1e01772a

See more details on using hashes here.

File details

Details for the file ai_factory_model-0.0.8-py3-none-any.whl.

File metadata

File hashes

Hashes for ai_factory_model-0.0.8-py3-none-any.whl
Algorithm Hash digest
SHA256 1cbfa86694a10a3d7415465ba8b5a0675f7d3cf134083e2e95bafbb1b4e81484
MD5 6540d94b6cc136bc092b17f79af3bc78
BLAKE2b-256 ef430b63bf36c0e6000c681f5c7e975167a1b70a27f031ca797663b45ba36472

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page