* **Machine Learning**: Normalmente exige muita multiplicação de matrizes e vetores. Pense numa grande planilha com números e em multiplicar todos eles juntos e ao mesmo tempo.

* **Deep Learning**: Esse é um subcampo do Machine Learning, então, o mesmo se aplica. A diferença é que não há apenas uma grande planilha com números para multiplicar, mas um grande conjunto delas, e em muitos casos, você utiliza um processador especial para construir e/ou usar esses modelos.

## Container Images { #container-images }

Docker has been one of the main tools to create and manage **container images** and **containers**.

And there's a public <a href="https://hub.docker.com/" class="external-link" target="_blank">Docker Hub</a> with pre-made **official container images** for many tools, environments, databases, and applications.

Por outro lado, existem milhares de usuários no chat, então tem uma grande chance de você encontrar alguém para trocar uma idéia por lá em qualquer horário. 😄

## Patrocine o autor

	AMCASH		R14, (R13), R12 // acb95838
	AMCASW		R14, (R13), R12 // ac395938
	AMCASV		R14, (R13), R12 // acb95938
	AMADDW		R14, (R13), R12 // ac396138
	AMADDV		R14, (R13), R12 // acb96138
	AMANDW		R14, (R13), R12 // ac396238
	AMANDV		R14, (R13), R12 // acb96238
	AMORW		R14, (R13), R12 // ac396338
	AMORV		R14, (R13), R12 // acb96338
	AMXORW		R14, (R13), R12 // ac396438
	AMXORV		R14, (R13), R12 // acb96438

        // Modify the original raw array
        raw[0] = 0x05;

        // Ensure the AvPair's internal raw array is not affected (defensive copy not made)
        // This test assumes that the AvPair constructor does NOT make a defensive copy of the raw array.
        // If a defensive copy were made, this test would fail, and the behavior would be more robust.

Once you are ready and want to use a tool to **manage the entire project**, packages dependencies, virtual environments, etc. I would suggest you try <a href="https://github.com/astral-sh/uv" class="external-link" target="_blank">uv</a>.

`uv` can do a lot of things, it can:

* **Install Python** for you, including different versions
* Manage the **virtual environment** for your projects
* Install **packages**

# FastAPI CLI { #fastapi-cli }

**FastAPI CLI** is a command line program that you can use to serve your FastAPI app, manage your FastAPI project, and more.

When you install FastAPI (e.g. with `pip install "fastapi[standard]"`), it includes a package called `fastapi-cli`, this package provides the `fastapi` command in the terminal.

To run your FastAPI app for development, you can use the `fastapi dev` command:

<div class="termy">

```console

* **Rápido para codar**: Aumenta a velocidade para desenvolver recursos entre 200% a 300%. *
* **Poucos bugs**: Reduz cerca de 40% de erros induzidos por humanos (desenvolvedores). *
* **Intuitivo**: Grande suporte a _IDEs_. <abbr title="também conhecido como _auto-complete_, _autocompletion_, _IntelliSense_">_Auto-Complete_</abbr> em todos os lugares. Menos tempo debugando.

As **FastAPI** is based on Starlette and implements the <abbr title="Asynchronous Server Gateway Interface">ASGI</abbr> specification, you can use any ASGI middleware.

A middleware doesn't have to be made for FastAPI or Starlette to work, as long as it follows the ASGI spec.

In general, ASGI middlewares are classes that expect to receive an ASGI app as the first argument.

     * Multiple calls to this method may be made concurrently.
     */
    @Throws(IOException::class)
    abstract fun onStream(stream: Http2Stream)

    /**
     * Notification that the connection's peer's settings may have changed to [settings].
     * Implementations should take appropriate action to handle the updated settings.
     *