altrumai 1.0.0

A python client for interacting with AltrumAI APIs.

AltrumAI Python API library

The AltrumAI Python library provides convenient access to the AltrumAI REST API from any Python 3.7+ application. The library includes type definitions for all request params and response fields, and offers synchronous and asynchronous client powered by requests.

Documentation

The REST API direct access and its documentation is still being worked upon.

Installation

[!IMPORTANT] The Altrum AI Python SDK is not public and is only available for internal aligne use or our customers (On request) at the moment.

pip install altrumai

Usage

import os
from altrumai import AltrumAI
from dotenv import load_dotenv

load_dotenv()

def get_models():

    # Initialise AltrumAI client
    client = AltrumAI(
        workspace=os.getenv("WORKSPACE_ID"),
        base_url=os.getenv("DEPLOYMENT_URL"),
        api_key=os.getenv("ALIGNEAI_API_KEY")
        )
    # Generate chat completion using AltrumAI
    response = client.models()

    # Log the response into your stdout
    print(response.text)

if __name__ == "__main__":
    get_models()

Async Usage

import os
from altrumai import AsyncAltrumAI
from dotenv import load_dotenv
import asyncio

load_dotenv()

async def get_models():

    # Initialise AltrumAI client
    client = AsyncAltrumAI(
        workspace=os.getenv("WORKSPACE_ID"),
        base_url=os.getenv("DEPLOYMENT_URL"),
        api_key=os.getenv("ALIGNEAI_API_KEY")
        )

    # Generate chat completion using AltrumAI
    response = await client.models()

    # Log the response into your stdout
    print(response.text)

if __name__ == "__main__":
    asyncio.run(get_models())

Models

List and describe the various models available in the API.

above usage code will give the list of all the available models

Paramenters : none

Deployments

List and describe the various models available in your workspace.

response = client.deployments()

Paramenters : none

Privacy

Detect entities as per data privacy compliance laws.

response = client.privacy(
        input="The UK company Steeper (http://bebionic.com/) is producing cutting edge electronic hands for amputees which feature over a dozen grip patterns, wrist mobility, and speed control - all through the muscle pressure sensors located in the device’s casing.",
        compliance= ["GDPR","HIPAA"],
        custom=["Money", "File Name"]
)

Parameters :

• input (str): The input text that needs to be checked for privacy compliance.
• compliance (list): A list of standard compliance regulations to check against. Supported values include: Supported values : ["GDPR" "HIPAA"]
• custom (list): A list of custom compliance rules to apply Supported values : ["Money" "Filename", "Account Number", "Password", "Marital Status", "Organization", "Dosage"]

Chat Completions

Given a list of messages comprising a conversation, the model will return a response.

response = client.chat_completion(
    model="mistral-7b-chat",
    messages=[{'role': 'user', 'content': "What are different types of emerging risks related to AI?"}],
    stream=False,
    timeout=60
)

Parameters :

• model (str): The name of the model to be used for generating responses(can get the list from client.models).
• messages (list): A list of message objects, each containing: role (str): The role of the message sender. Valid roles are: "user" "assistant" content (str): The content of the message.
• stream (bool, optional): If set to True, the response will be streamed as it is generated. Defaults to False.

Moderations

Detect toxicity and bias for moderations.

response = client.moderations(
        input="The UK company Steeper (http://bebionic.com/) is producing cutting edge electronic hands for amputees which feature over a dozen grip patterns, wrist mobility, and speed control - all through the muscle pressure sensors located in the device’s casing.",
        guardrails= ["bias","toxicity"]
)

Parameters :

• input (str): The input text that needs to be moderated.
• guardrails (list): A list of guardrails to check the input text against. Supported values include: "bias" "toxicity"

Embeddings

Creates an embedding vector representing the input text.

response = client.create_embeddings(
    inputs=["The new coffee blend from BeanWorks is robust and flavorful"],
    model="nomic-embed-v1.5",
    dimensions=64,
    encoding_format="float",
)

Parameters :

• inputs (list): A list of input texts for which embeddings need to be generated.
• model (str): The name of the model to be used for generating embeddings(can get the list from client.models).
• dimensions (int): The number of dimensions for the generated embeddings. Supported values : 64, 128, 256, 512, 768
• encoding_format (str): The format of the embedding values. Supported value: "float".

Ping

Endpoint to verify your Workspace ID and API Key on AltrumAI

ping_response = client.ping()

Timeouts

Currently by default 60 sec timeout is set for post methods, but timeout can be passed as a parameter while calling any method.

response = client.deployments(timeout=60)

Module-level client

We recommend that you always instantiate a client (e.g., with client = AltrumAI()) in application code.

Handling errors

When the library is unable to connect to the API (for example, due to network connection problems or a timeout), a subclass of altrumai.APIConnectionError will be raised.

When the API returns a non-success status code (that is, 4xx or 5xx response), a subclass of altrumai.APIStatusError is raised, containing status_code and response properties.

All errors inherit from altrumai.APIError.

Error codes are as followed:

Status Code	Error Type
400	BadRequestError
401	AuthenticationError
403	PermissionDeniedError
404	NotFoundError
422	UnprocessableEntityError
429	RateLimitError
>=500	InternalServerError
N/A	APIConnectionError

(Work In Progress)

Retries

Certain errors are automatically retried 2 times by default, with a short exponential backoff. Connection errors (for example, due to a network connectivity problem), 408 Request Timeout, 409 Conflict, 429 Rate Limit, and >=500 Internal errors are all retried by default.

You can use the max_retries option to configure or disable retry settings. (Work In Progress)

Versioning

This package generally follows SemVer conventions, though certain backwards-incompatible changes may be released as minor versions:

1. Changes that only affect static types, without breaking runtime behavior.
2. Changes to library internals which are technically public but not intended or documented for external use. (Please open a GitHub issue to let us know if you are relying on such internals).
3. Changes that we do not expect to impact the vast majority of users in practice.

We take backwards-compatibility seriously and work hard to ensure you can rely on a smooth upgrade experience.

We are keen for your feedback; please open an issue with questions, bugs, or suggestions.