Xorbits Inference: Model Serving Made Easy 🤖

English | 中文介绍 | 日本語

Xorbits Inference(Xinference) is a powerful and versatile library designed to serve language, speech recognition, and multimodal models. With Xorbits Inference, you can effortlessly deploy and serve your or state-of-the-art built-in models using just a single command. Whether you are a researcher, developer, or data scientist, Xorbits Inference empowers you to unleash the full potential of cutting-edge AI models.

🌟 Model Serving Made Easy: Simplify the process of serving large language, speech recognition, and multimodal models. You can set up and deploy your models for experimentation and production with a single command.

⚡️ State-of-the-Art Models: Experiment with cutting-edge built-in models using a single command. Inference provides access to state-of-the-art open-source models!

🖥 Heterogeneous Hardware Utilization: Make the most of your hardware resources with ggml. Xorbits Inference intelligently utilizes heterogeneous hardware, including GPUs and CPUs, to accelerate your model inference tasks.

⚙️ Flexible API and Interfaces: Offer multiple interfaces for interacting with your models, supporting RPC, RESTful API(compatible with OpenAI API), CLI and WebUI for seamless management and monitoring.

🌐 Distributed Deployment: Excel in distributed deployment scenarios, allowing the seamless distribution of model inference across multiple devices or machines.

🔌 Built-in Integration with Third-Party Libraries: Xorbits Inference seamlessly integrates with popular third-party libraries like LangChain and LlamaIndex. (Coming soon)

Xinference can be installed via pip from PyPI. It is highly recommended to create a new virtual environment to avoid conflicts.

$ pip install "xinference"

xinference installs basic packages for serving models.

To serve ggml models, you need to install the following extra dependencies:

$ pip install "xinference[ggml]"

If you want to achieve acceleration on different hardware, refer to the installation documentation of the corresponding package.

• llama-cpp-python is required to run baichuan, wizardlm-v1.0, vicuna-v1.3 and orca.
• chatglm-cpp-python is required to run chatglm and chatglm2.

To serve PyTorch models, you need to install the following extra dependencies:

$ pip install "xinference[pytorch]"

If you want to serve all the supported models, install all the dependencies:

$ pip install "xinference[all]"

You can deploy Xinference locally with a single command or deploy it in a distributed cluster.

To start a local instance of Xinference, run the following command:

$ xinference

To deploy Xinference in a cluster, you need to start a Xinference supervisor on one server and Xinference workers on the other servers. Follow the steps below:

Starting the Supervisor: On the server where you want to run the Xinference supervisor, run the following command:

$ xinference-supervisor -H "${supervisor_host}"

Replace ${supervisor_host} with the actual host of your supervisor server.

Starting the Workers: On each of the other servers where you want to run Xinference workers, run the following command:

$ xinference-worker -e "http://${supervisor_host}:9997"

Once Xinference is running, an endpoint will be accessible for model management via CLI or Xinference client.

• For local deployment, the endpoint will be http://localhost:9997.
• For cluster deployment, the endpoint will be http://${supervisor_host}:9997, where ${supervisor_host} is the hostname or IP address of the server where the supervisor is running.

You can also view a web UI using the Xinference endpoint to chat with all the builtin models. You can even chat with two cutting-edge AI models side-by-side to compare their performance!

Xinference provides a command line interface (CLI) for model management. Here are some useful commands:

• Launch a model (a model UID will be returned): xinference launch
• List running models: xinference list
• List all the builtin models: xinference list --all
• Terminate a model: xinference terminate --model-uid ${model_uid}

Xinference also provides a client for managing and accessing models programmatically:

from xinference.client import Client

client = Client("http://localhost:9997")
model_uid = client.launch_model(model_name="chatglm2")
model = client.get_model(model_uid)

chat_history = []
prompt = "What is the largest animal?"
model.chat(
            prompt,
            chat_history,
            generate_config={"max_tokens": 1024}
        )

Result:

{
  "id": "chatcmpl-8d76b65a-bad0-42ef-912d-4a0533d90d61",
  "model": "56f69622-1e73-11ee-a3bd-9af9f16816c6",
  "object": "chat.completion",
  "created": 1688919187,
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "The largest animal that has been scientifically measured is the blue whale, which has a maximum length of around 23 meters (75 feet) for adult animals and can weigh up to 150,000 pounds (68,000 kg). However, it is important to note that this is just an estimate and that the largest animal known to science may be larger still. Some scientists believe that the largest animals may not have a clear \"size\" in the same way that humans do, as their size can vary depending on the environment and the stage of their life."
      },
      "finish_reason": "None"
    }
  ],
  "usage": {
    "prompt_tokens": -1,
    "completion_tokens": -1,
    "total_tokens": -1
  }
}

See examples for more examples.

To view the builtin models, run the following command:

$ xinference list --all

NOTE:

• Xinference will download models automatically for you, and by default the models will be saved under ${USER}/.xinference/cache.
• Foundation models only provide interface generate.
• RLHF and SFT models provide both generate and chat.
• If you want to use Apple Metal GPU for acceleration, please choose the q4_0 and q4_1 quantization methods.
• llama-2-chat 70B ggmlv3 model only supports q4_0 quantization currently.

Pytorch has been integrated recently, and the usage scenarios are described below:

• Foundation Model：baichuan（7B、13B）。
• SFT Model：baichuan-chat（13B）、vicuna-v1.3（7B、13B、33B）。

• CUDA: On Linux and Windows systems, cuda device is used by default.
• MPS: On Mac M1/M2 devices, mps device is used by default.
• CPU: It is not recommended to use a cpu device, as it takes up a lot of memory and the inference speed is very slow.

• none: indicates that no quantization is used.
• 8-bit: use 8-bit quantization.
• 4-bit: use 4-bit quantization. Note: 4-bit quantization is only supported on Linux systems and CUDA devices.

• On MacOS system, baichuan-chat model is not supported, and baichuan model cannot use 8-bit quantization.

The table below shows memory usage and supported devices of some models.

Xinference is currently under active development. Here's a roadmap outlining our planned developments for the next few weeks:

With Xinference, it will be much easier for users to use these libraries and build applications with LLMs.