Skip to main content

LLM Adapters Package Documentation

List of currently supported models

Overview

The Adapters package facilitates communication between different language model APIs by providing a unified interface for interaction. This ensures ease of use and flexibility in integrating multiple models from various providers.

The package can be installed an used via pip:

pip install martian-adapters

Getting Started

Prerequisites

  • Python version: 3.11.10
  • Poetry

Installation

poetry install
poetry run pre-commit install

Setting Up Pre-commit

To run pre-commit manually:

poetry run pre-commit run --all-files

Semantic Versioning

For versioning we follow Semantic Versioning

Environment Configuration

The package requires certain environment variables to be set by the users:

  • Copy .env-example to .env and populate it with appropriate values.

Running Tests

poetry run pytest

Quickstart

from adapters import AdapterFactory

# First component in model path is Provider, then Vendor, and last model name itself
adapter = AdapterFactory.get_adapter_by_path("openai/openai/gpt-4o-mini")

adapter.execute_sync(
    [
        {"role": "system", "content": "You are a helpful assistant."},
        {
            "role": "user",
            "content": "Write a haiku about recursion in programming.",
        },
    ]
)

Adapter paths follows the format provider/vendor/model_name. Use AdapterFactory.get_supported_models() to retrieve all supported models. For a given model, model.get_path() returns the adapter path.

Contributing

Adding New Models

  1. Existing Providers: Add new models to the MODELS array if the provider is already supported.

  2. New Providers:

    • If the provider follows the OpenAI format, model integration is straightforward. See the "Fireworks" provider class as an example.
    • For providers with different schemas, see the "Anthropic" provider class for guidance.

Development Steps

  1. Add the Provider and Model: Update provider_adapters/__init__.py and test files accordingly.

  2. Write Tests: Add tests in the relevant directories. Use @pytest.mark.vcr for tests making network requests.

  3. Run Tests:

    poetry run pytest
    
  4. Check-in Cassette Files: Include any new cassette YAML files in your commit.

  5. Send a Pull Request: Ensure all tests pass before requesting a review.

Re-creating Cassette Files

Use the poetry run pytest --record-mode=rewrite option with pytest to update cassette files.

Additional Notes

Some models may only be accessible from specific locations (e.g., the U.S.). In such cases, running tests might require access to a U.S.-based server.

This documentation provides a streamlined approach to using and contributing to the Adapters package, emphasizing practical steps and clear examples.

Misclenous

HTTP Client configuration

To optimize throughput and performance, we provide options to configure HTTP networking parameters:

ADAPTERS_MAX_KEEPALIVE_CONNECTIONS_PER_PROCESS = 100
ADAPTERS_MAX_CONNECTIONS_PER_PROCESS = 1000
ADAPTERS_HTTP_CONNECT_TIMEOUT = 5
ADAPTERS_HTTP_TIMEOUT = 600

Base URL overriding

For stress testing or other purposes, you can override all base URLs by setting the following in your .env file:

_ADAPTERS_OVERRIDE_ALL_BASE_URLS_ = "https://new-base-url.com/api"

This setting ensures that all LLM API calls will route to the specified new base URL.

Metadata

Release files for martian-adapters 7.25.0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for martian-adapters 7.25.0
File Size Uploaded Provenance Actions
martian_adapters-7.25.0.tar.gz 39.8 kB Without provenance Details for martian_adapters-7.25.0.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for martian-adapters 7.25.0
File Interpreter ABI Platform Actions
martian_adapters-7.25.0-py3-none-any.whl Python 3 none any Details for martian_adapters-7.25.0-py3-none-any.whl

Total release size: 98.6 kB

Release files / martian_adapters-7.25.0.tar.gz

Download URL martian_adapters-7.25.0.tar.gz
Size 39.8 kB
Tags Source
SHA-256 checksum
How to use checksums
d97697548c69563d11ec8490ea0199461b748c6f509b25d3c422948c4e2f2ab7
BLAKE2b-256 checksum
How to use checksums
b2e8db393eaac9daf35d15e4c8538539dcf1241d30ec0607d4c31236b150c31c
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via poetry/2.1.3 CPython/3.13.5 Darwin/24.5.0

Release files / martian_adapters-7.25.0-py3-none-any.whl

Download URL martian_adapters-7.25.0-py3-none-any.whl
Size 58.9 kB
Tags Python 3
SHA-256 checksum
How to use checksums
f9b324c29b4ee4fe04580411a7db9922dd139efaca31be6da9a89cf009fc2b4d
BLAKE2b-256 checksum
How to use checksums
3a4cef3b9231ce1a9dc34b8b97c533108bb6bdfc24b9d962d753ece1e3a9ba56
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via poetry/2.1.3 CPython/3.13.5 Darwin/24.5.0

Security for martian-adapters 7.25.0

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Release history Release notifications | RSS feed

This release

7.25.0 This release

2 release files No provenance

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page