Skip to main content

('A library for augmenting large language models using MLX',)

Project description

mlx_augllm

MLX(Apple Silicon向け機械学習フレームワーク)を用いた、 ローカルで動作する LLM / VLM のための統一インターフェースライブラリです。

本ライブラリは以下を目的としています。

  • ローカルLLM/VLMを簡単かつ一貫したAPIで扱える
  • Tool Use(関数呼び出し)に対応
  • 会話履歴の管理を自動化
  • Apple Siliconに最適化

インストール

pip install -U mlx_augllm
  • Apple Silicon必須

サンプル

from mlx_augllm import MlxAugmentedLLM, MlxLLMInterface, PromptBuilder 

def run_test():

    # モデルの準備
    model_path = "mlx-community/gemma-3-27b-it-4bit"
    augmented_llm = MlxAugmentedLLM(
        llm_interface=MlxLLMInterface(
            model_path=model_path,
            use_vision=False,
            temp=0.7,
            top_k=50,
            top_p=0.9,
            min_p=0.05,
            max_tokens=8192
        ),
        prompt_builder=PromptBuilder(system_prompt_text="あなたは有能なアシスタントです。"),
    )

    # 実行テスト
    user_query = "トポロジー最適化について教えてください。"
    
    print(f"\nユーザーの問いかけ: {user_query}")
    print("-" * 50)
    print("AIの応答 (Streaming):")

    # respond の呼び出し (contextを渡す)
    response_generator = augmented_llm.respond(
        user_text=user_query,
        stream=True,
        temp=0.7
    )

    full_response = ""
    for chunk in response_generator:
        print(chunk, end="", flush=True)
        full_response += chunk
    
    print("\n" + "-" * 50)
    print("【内部レポート】")
    if augmented_llm.report_text:
        print(f"最終回答の文字数: {len(augmented_llm.report_text)}")

if __name__ == "__main__":
    run_test()

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

mlx_augllm-1.1.tar.gz (17.6 kB view details)

Uploaded Source

Built Distribution

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

mlx_augllm-1.1-py3-none-any.whl (19.9 kB view details)

Uploaded Python 3

File details

Details for the file mlx_augllm-1.1.tar.gz.

File metadata

  • Download URL: mlx_augllm-1.1.tar.gz
  • Upload date:
  • Size: 17.6 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.13.9

File hashes

Hashes for mlx_augllm-1.1.tar.gz
Algorithm Hash digest
SHA256 631406dfb39d5e7323174f195ec98786ab47618448752b631125b8338ae2f0ab
MD5 1f7e6b5fdf51b4cf6952ffd3b6d62af6
BLAKE2b-256 c72d57eacba4e91b13bbce7003b24577afd9478c270c7838907165e6274d52d3

See more details on using hashes here.

File details

Details for the file mlx_augllm-1.1-py3-none-any.whl.

File metadata

  • Download URL: mlx_augllm-1.1-py3-none-any.whl
  • Upload date:
  • Size: 19.9 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.13.9

File hashes

Hashes for mlx_augllm-1.1-py3-none-any.whl
Algorithm Hash digest
SHA256 5cbfe59d60b78bb57cea3b5f900e0cc642fcf9043e47fd15c3e341f79b9b481e
MD5 88a9d089c2cb8efe6c063bd783eeec8a
BLAKE2b-256 6e172d25ac6a585db0f032e259ab22971932d45185276b28835cd45f534d6610

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