...会在响应把慢速数据发送完之后才运行：

{* ../../docs_src/dependencies/tutorial013_an_py310.py ln[30:38] hl[31:33] *}

但由于 `generate_stream()` 并不使用数据库会话，因此在发送响应期间保持会话打开并非必要。

如果你使用的是 SQLModel（或 SQLAlchemy）并碰到这种特定用例，你可以在不再需要时显式关闭会话：

{* ../../docs_src/dependencies/tutorial014_an_py310.py ln[24:28] hl[28] *}

这样会话会释放数据库连接，让其他请求可以使用。

...會在回應完成傳送這些慢速資料後才執行：

{* ../../docs_src/dependencies/tutorial013_an_py310.py ln[30:38] hl[31:33] *}

但因為 `generate_stream()` 並未使用資料庫 session，實際上不需要在傳送回應時保持 session 開啟。

如果你用的是 SQLModel（或 SQLAlchemy）且有這種特定情境，你可以在不再需要時明確關閉該 session：

{* ../../docs_src/dependencies/tutorial014_an_py310.py ln[24:28] hl[28] *}

如此一來，該 session 就會釋放資料庫連線，讓其他請求可以使用。

## 0.68.1 (2021-08-24)

* ✨ Add support for `read_with_orm_mode`, to support [SQLModel](https://sqlmodel.tiangolo.com/) relationship attributes. PR [#3757](https://github.com/tiangolo/fastapi/pull/3757) by [@tiangolo](https://github.com/tiangolo).

### Translations

from fastapi import FastAPI
from fastapi.testclient import TestClient
from pydantic import BaseModel

app = FastAPI()


class SubModel(BaseModel):
    a: str | None = "foo"


class Model(BaseModel):
    x: int | None = None
    sub: SubModel


class ModelSubclass(Model):
    y: int
    z: int = 0
    w: int | None = None


class ModelDefaults(BaseModel):
    w: str | None = None
    x: str | None = None

    }

    @Test
    public void test_setModel() {
        final LlmChatResponse response = new LlmChatResponse();

        response.setModel("llama3");
        assertEquals("llama3", response.getModel());

        response.setModel("gpt-3.5-turbo");
        assertEquals("gpt-3.5-turbo", response.getModel());
    }

    @Test
    public void test_nullValues() {

All that, arbitrarily nested.

### Define a submodel { #define-a-submodel }

For example, we can define an `Image` model:

{* ../../docs_src/body_nested_models/tutorial004_py310.py hl[7:9] *}

### Use the submodel as a type { #use-the-submodel-as-a-type }

And then we can use it as the type of an attribute:

var (
	mysqlDSN     = "gorm:gorm@tcp(localhost:9910)/gorm?charset=utf8&parseTime=True&loc=Local"
	postgresDSN  = "user=gorm password=gorm dbname=gorm host=localhost port=9920 sslmode=disable TimeZone=Asia/Shanghai"
	gaussdbDSN   = "user=gaussdb password=Gaussdb@123 dbname=gorm host=localhost port=9950 sslmode=disable TimeZone=Asia/Shanghai"
	sqlserverDSN = "sqlserver://sa:LoremIpsum86@localhost:9930?database=master"

    }

    @Test
    public void test_setModel() {
        final LlmChatRequest request = new LlmChatRequest();

        final LlmChatRequest result = request.setModel("gpt-4");

        assertSame(request, result);
        assertEquals("gpt-4", request.getModel());
    }

    @Test
    public void test_setMaxTokens() {

Hem de istediğiniz kadar iç içe.

### Bir alt model tanımlayın { #define-a-submodel }

Örneğin bir `Image` modeli tanımlayabiliriz:

{* ../../docs_src/body_nested_models/tutorial004_py310.py hl[7:9] *}

### Alt modeli tip olarak kullanın { #use-the-submodel-as-a-type }

Ardından bunu bir attribute’un tipi olarak kullanabiliriz:

                },
                "/response_model_filtering_model-annotation_submodel-return_submodel": {
                    "get": {
                        "summary": "Response Model Filtering Model Annotation Submodel Return Submodel",
                        "operationId": "response_model_filtering_model_annotation_submodel_return_submodel_response_model_filtering_model_annotation_submodel_return_submodel_get",
                        "responses": {