"publishPluginsToTestRepository",
        ) -> false

        // Core tasks
        task.name in listOf(
            "components",
            "dependantComponents",
            "model",
        ) -> true

        task.name.startsWithAnyOf(
            "publish",
            "idea",
        ) -> true

        task is GradleBuild -> true

        // gradle/gradle build tasks

首先，确保你创建并激活了[虚拟环境](../virtual-environments.md)，然后安装 `sqlmodel`：

<div class="termy">

```console
$ pip install sqlmodel
---> 100%
```

</div>

## 创建含有单一模型的应用 { #create-the-app-with-a-single-model }

我们先创建应用的最简单的第一个版本，只有一个 **SQLModel** 模型。

稍后我们将通过下面的**多个模型**提高其安全性和多功能性。🤓

### 创建模型 { #create-models }

导入 `SQLModel` 并创建一个数据库模型：

	//  a large number of policies
	r, err := ies.client.Get(ctx, iamConfigPoliciesPrefix, etcd.WithPrefix())
	if err != nil {
		return err
	}

	// Parse all values to construct the policies data model.
	for _, kvs := range r.Kvs {
		if err = ies.getPolicyDocKV(ctx, kvs, m); err != nil && !errors.Is(err, errNoSuchPolicy) {
			return err
		}
	}
	return nil
}

 * For 4.0.0 models: applies limited inference (parent-related only).
 * For 4.1.0+ models: applies full inference optimizations.
 * Removes redundant information that can be inferred by Maven during model building.
 */
@Named
@Singleton
@Priority(30)
public class InferenceStrategy extends AbstractUpgradeStrategy {

    @Override
    public boolean isApplicable(UpgradeContext context) {

    public static final String ERROR_TIMEOUT = "timeout";

    /** Error code for context length exceeded. */
    public static final String ERROR_CONTEXT_LENGTH_EXCEEDED = "context_length_exceeded";

    /** Error code for model not found. */
    public static final String ERROR_MODEL_NOT_FOUND = "model_not_found";

    /** Error code for invalid response from the LLM provider. */
    public static final String ERROR_INVALID_RESPONSE = "invalid_response";

      description: If compiling from source
  - type: input
    id: Cuda
    attributes:
      label: CUDA/cuDNN version
  - type: input
    id: Gpu
    attributes:
      label: GPU model and memory
      description: If compiling from source
  - type: textarea
    id: what-happened
    attributes:
      label: Current behavior?
      description: Also tell us, what did you expect to happen?

首先，从 Pydantic 中导入 `Field`：

{* ../../docs_src/body_fields/tutorial001_an_py310.py hl[4] *}

/// warning | 警告

注意，与从 `fastapi` 导入 `Query`，`Path`、`Body` 不同，要直接从 `pydantic` 导入 `Field` 。

///

## 声明模型属性 { #declare-model-attributes }

然后，使用 `Field` 定义模型的属性：

{* ../../docs_src/body_fields/tutorial001_an_py310.py hl[11:14] *}

`Field` 的工作方式和 `Query`、`Path`、`Body` 相同，参数也相同。

/// note | 技术细节

### Consistency Guarantees

MinIO follows strict **read-after-write** and **list-after-write** consistency model for all i/o operations both in distributed and standalone modes. This consistency model is only guaranteed if you use disk filesystems such as xfs, zfs or btrfs etc.. for distributed setup.

   * triggered by cancel(). For details, see the comments on the fields of TimeoutFuture.
   *
   * <p>{@code volatile} is required for j2objc transpiling:
   * https://developers.google.com/j2objc/guides/j2objc-memory-model#atomicity
   */
  private volatile @Nullable InterruptibleTask<?> task;

  TrustedListenableFutureTask(Callable<V> callable) {
    this.task = new TrustedFutureInterruptibleTask(callable);
  }

    </ul>
    ]]>
  </description>
  <defaults>
    <default>
      <key>package</key>
      <value>org.apache.maven.model</value>
    </default>
  </defaults>
  <classes>
    <class rootElement="true" xml.tagName="project">
      <name>Model</name>
      <superClass>ModelBase</superClass>
      <description>
        The {@code &lt;project>} element is the root of the descriptor.