* **Aprendizado de Máquina**: 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 Aprendizado de Máquina, 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.

O resultado de chamar a função é algo que pode ser codificado com o padrão do Python <a href="https://docs.python.org/3/library/json.html#json.dumps" class="external-link" target="_blank">`json.dumps()`</a>.

A função não retorna um grande `str` contendo os dados no formato JSON (como uma string). Mas sim, retorna uma estrutura de dados padrão do Python (por exemplo, um `dict`) com valores e subvalores compatíveis com JSON.

/// note | Nota

  mdx_include: null
  markdown_include_variants: null
extra:
  social:
  - icon: fontawesome/brands/github-alt
    link: https://github.com/fastapi/fastapi
  - icon: fontawesome/brands/discord
    link: https://discord.gg/VQjSZaeJmf
  - icon: fontawesome/brands/twitter
    link: https://x.com/fastapi
  - icon: fontawesome/brands/linkedin
    link: https://www.linkedin.com/company/fastapi
  - icon: fontawesome/solid/globe

= vendor/github.com/beorn7/perks licensed under: =

Copyright (C) 2013 Blake Mizerany

Permission is hereby granted, free of charge, to any person obtaining
a copy of this software and associated documentation files (the
"Software"), to deal in the Software without restriction, including
without limitation the rights to use, copy, modify, merge, publish,
distribute, sublicense, and/or sell copies of the Software, and to

The MIT License (MIT)

Copyright (c) 2018 Sebastián Ramírez

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is

        .initializer(
          """
          |%T(⇥
          |sections = "%L",
          |ranges = "%L",
          |mappings = "%L",
          |⇤)
          """.trimMargin(),
          idnaMappingTable,
          data.sections.escapeDataString(),
          data.ranges.escapeDataString(),
          data.mappings.escapeDataString(),
        ).build(),
    ).build()
}

/**

= vendor/github.com/MakeNowJust/heredoc licensed under: =

The MIT License (MIT)

Copyright (c) 2014-2019 TSUYUSATO Kitsune

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell

= vendor/github.com/blang/semver/v4 licensed under: =

The MIT License

Copyright (c) 2014 Benedikt Lang <github at benediktlang.de>

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell

= vendor/github.com/cespare/xxhash/v2 licensed under: =

Copyright (c) 2016 Caleb Spare

MIT License

Permission is hereby granted, free of charge, to any person obtaining
a copy of this software and associated documentation files (the
"Software"), to deal in the Software without restriction, including
without limitation the rights to use, copy, modify, merge, publish,
distribute, sublicense, and/or sell copies of the Software, and to

   * request, and ensure that (1/QPS) seconds have elapsed since then. For example, for a rate of
   * QPS=5 (5 tokens per second), if we ensure that a request isn't granted earlier than 200ms after
   * the last one, then we achieve the intended rate. If a request comes and the last request was