Utils

check_llm_as_a_judge_client()

Check the LLM as a judge client by sending a message to the model.

Uses Langchain OpenAIChat ou Langchain AzureChatas the LLM client

Source code in src/utils.py

def check_llm_as_a_judge_client():
    """Check the LLM as a judge client by sending a message to the model.

    Uses Langchain OpenAIChat ou Langchain AzureChatas the LLM client
    """
    try:
        llmaaj_chat_client, llmaaj_client_embedding = get_llm_as_a_judge_client()

        messages = [
            (
                "system",
                "You are a helpful assistant that translates English to French. Translate the user sentence.",
            ),
            ("human", "I love programming."),
        ]
        ai_msg = llmaaj_chat_client.invoke(messages)
        logger.info(
            f"\nEvaluation environment variables are: \n{pretty_repr(settings.get_eval_env_vars())}\n"
            f"\nmodel response: {ai_msg.content}"
        )

    except Exception as e:
        logger.error(
            f"Error in check_llm_as_a_judge_client:"
            f"\nevaluation environment variables are: {pretty_repr(settings.get_eval_env_vars())}"
        )
        raise e

check_llm_client()

Check the LLM client by sending a message to the model.

Uses OpenAI/AzureOpenAI as the LLM client

Source code in src/utils.py

def check_llm_client():
    """Check the LLM client by sending a message to the model.

    Uses OpenAI/AzureOpenAI as the LLM client
    """
    client, model_name = get_llm_client()
    try:
        chat_completion = client.chat.completions.create(
            messages=[
                {
                    "role": "user",
                    "content": "Hi",
                }
            ],
            model=model_name,
        )
        logger.info(chat_completion)

        logger.info(
            f"\nActive environment variables are: \n{pretty_repr(settings.get_active_env_vars())}\n"
            f"\nmodel response: {chat_completion.choices[0].message.content}"
        )
    except Exception as e:
        logger.error(
            f"Error in check_llm_client:"
            f"\nActive environment variables are: {pretty_repr(settings.get_active_env_vars())}"
        )
        raise e

get_llm_as_a_judge_client()

Initializes and returns a LLM as a judge client based on the configured provider.

Depending on the value of settings.LLMAAJ_PROVIDER, this function will initialize either an OpenAI or AzureOpenAI client and embedding client from Langchain OpenAI library. It loads the client with the appropriate settings and logs the model being used.

If settings.ENABLE_EVALUATION is False, the function will return (None, None) and log a warning.

Returns:

Name	Type	Description
tuple		A tuple containing - the initialized client and the embedding client (AzureOpenAI and AzureOpenAIEmbeddings or OpenAI and OpenAIEmbeddings) - or (None, None) if evaluation is disabled.

Raises:

Type	Description
ValueError	If the configured LLM provider is unsupported.

Source code in src/utils.py

def get_llm_as_a_judge_client():
    """Initializes and returns a LLM as a judge client based on the configured provider.

    Depending on the value of `settings.LLMAAJ_PROVIDER`, this function will initialize
    either an OpenAI or AzureOpenAI client and embedding client from Langchain OpenAI library. It loads the client with
    the appropriate settings and logs the model being used.

    If `settings.ENABLE_EVALUATION` is False, the function will return `(None, None)` and log a warning.

    Returns:
        tuple: A tuple containing
            - the initialized client and the embedding client (AzureOpenAI and AzureOpenAIEmbeddings or OpenAI and
            OpenAIEmbeddings)
            - or `(None, None)` if evaluation is disabled.

    Raises:
        ValueError: If the configured LLM provider is unsupported.
    """
    if settings.ENABLE_EVALUATION:
        if settings.LLMAAJ_PROVIDER == ProviderEnum.azure_openai:
            client = AzureChatOpenAI(
                azure_endpoint=settings.LLMAAJ_AZURE_OPENAI_BASE_URL,
                deployment_name=settings.LLMAAJ_AZURE_OPENAI_DEPLOYMENT_NAME,
                openai_api_key=settings.LLMAAJ_AZURE_OPENAI_API_KEY,
                openai_api_version=settings.LLMAAJ_AZURE_OPENAI_API_VERSION,
            )
            embeddings_client = AzureOpenAIEmbeddings(
                azure_endpoint=settings.LLMAAJ_AZURE_OPENAI_BASE_URL,
                model=settings.LLMAAJ_AZURE_OPENAI_EMBEDDING_DEPLOYMENT_NAME,
                openai_api_key=settings.LLMAAJ_AZURE_OPENAI_API_KEY,
                openai_api_version=settings.LLMAAJ_AZURE_OPENAI_API_VERSION,
            )
            loguru_logger.info(
                f"Loaded LLMAAJ AzureOpenAI client with model: {settings.LLMAAJ_OPENAI_DEPLOYMENT_NAME}"
            )
        elif settings.LLMAAJ_PROVIDER == ProviderEnum.openai:
            client = ChatOpenAI(
                model_name=settings.LLMAAJ_OPENAI_DEPLOYMENT_NAME,
                openai_api_base=settings.LLMAAJ_OPENAI_BASE_URL,
                openai_api_key=settings.LLMAAJ_OPENAI_API_KEY,
            )
            embeddings_client = OpenAIEmbeddings(
                model=settings.LLMAAJ_OPENAI_EMBEDDING_DEPLOYMENT_NAME,
                openai_api_base=settings.LLMAAJ_OPENAI_BASE_URL,
                openai_api_key=settings.LLMAAJ_OPENAI_API_KEY,
            )
            loguru_logger.info(
                f"Loaded LLMAAJ OpenAI client with model: {settings.LLMAAJ_OPENAI_DEPLOYMENT_NAME}"
            )

        else:
            raise ValueError(f"Unsupported LLM provider: {settings.LLMAAJ_PROVIDER}")

        return (
            client,
            embeddings_client,
        )
    else:
        loguru_logger.warning(
            "Evaluation is disabled. Set ENABLE_EVALUATION to True to activate it."
        )
        return None, None

get_llm_client()

Initializes and returns a language model client based on the configured provider.

Depending on the value of settings.LLM_PROVIDER, this function will initialize either an OpenAI or AzureOpenAI client from OpenAI library. It loads the client with the appropriate settings and logs the model being used.

Returns:

Name	Type	Description
tuple	tuple[object, str]	A tuple containing the initialized client and the model name.

Raises:

Type	Description
ValueError	If the configured LLM provider is unsupported.

Source code in src/utils.py

def get_llm_client() -> tuple[object, str]:
    """Initializes and returns a language model client based on the configured provider.

    Depending on the value of `settings.LLM_PROVIDER`, this function will initialize
    either an OpenAI or AzureOpenAI client from OpenAI library. It loads the client with the appropriate
    settings and logs the model being used.

    Returns:
        tuple: A tuple containing the initialized client and the model name.

    Raises:
        ValueError: If the configured LLM provider is unsupported.
    """
    if settings.LLM_PROVIDER == ProviderEnum.openai:
        from openai import OpenAI

        client = OpenAI(
            base_url=settings.OPENAI_BASE_URL,
            api_key=settings.OPENAI_API_KEY.get_secret_value(),
        )
        model_name = settings.OPENAI_DEPLOYMENT_NAME
        loguru_logger.info(f"Loaded OpenAI client with model: {model_name}")

    elif settings.LLM_PROVIDER == ProviderEnum.azure_openai:
        from openai import AzureOpenAI

        client = AzureOpenAI(
            api_key=settings.AZURE_OPENAI_API_KEY.get_secret_value(),
            api_version=settings.AZURE_OPENAI_API_VERSION,
            azure_endpoint=settings.AZURE_OPENAI_BASE_URL,
        )
        model_name = settings.AZURE_OPENAI_DEPLOYMENT_NAME
        loguru_logger.info(f"Loaded AzureOpenAI client with model: {model_name}")

    else:
        raise ValueError(f"Unsupported LLM provider: {settings.LLM_PROVIDER}")

    return client, model_name

initialize()

Initialize the settings, logger, and search client.

Reads the environment variables from the .env file defined in the Settings class.

Returns:

Type	Description
	settings
	loguru_logger
	search_client

Source code in src/utils.py

def initialize():
    """Initialize the settings, logger, and search client.

    Reads the environment variables from the .env file defined in the Settings class.

    Returns:
        settings
        loguru_logger
        search_client
    """
    settings = Settings()
    loguru_logger.remove()

    if settings.DEV_MODE:
        loguru_logger.add(sys.stderr, level="TRACE")
    else:
        loguru_logger.add(sys.stderr, level="INFO")

    search_client = None
    if settings.ENABLE_AZURE_SEARCH:
        search_client = SearchClient(
            settings.AZURE_SEARCH_SERVICE_ENDPOINT,
            settings.AZURE_SEARCH_INDEX_NAME,
            AzureKeyCredential(settings.AZURE_SEARCH_API_KEY),
        )

    return settings, loguru_logger, search_client