TFE_Context* ctx_1 =
      CreateContext(serialized_server_def_1,
                    /*isolate_session_state=*/false, init_timeout_in_ms);

  // Remote device on `worker2`.
  const char remote_device[] = "/job:localhost/replica:0/task:2/device:CPU:0";
  // `ctx_0`, `ctx_1` contains `remote_device`.
  {
    const std::vector<std::string>& device_names = ListDeviceNames(ctx_0);

* Backporting TPR fix to 1.4 ([#42380](https://github.com/kubernetes/kubernetes/pull/42380), [@foxish](https://github.com/foxish))
* Fix AWS device allocator to only use valid device names ([#41455](https://github.com/kubernetes/kubernetes/pull/41455), [@gnufied](https://github.com/gnufied))

   * Returns a list that applies {@code function} to each element of {@code fromList}. The returned
   * list is a transformed view of {@code fromList}; changes to {@code fromList} will be reflected
   * in the returned list and vice versa.
   *
   * <p>Since functions are not reversible, the transform is one-way and new items cannot be stored
   * in the returned list. The {@code add}, {@code addAll} and {@code set} methods are unsupported

   *       instances.
   *   <li>For the constructor or static factory method used to construct instances, it's checked
   *       that when equal parameters are passed, the result instance should also be equal; and vice
   *       versa.
   *   <li>Inequality check is not performed against state mutation methods such as {@link
   *       List#add}, or functional update methods such as {@link

* Till this release, iscsi plugin was waiting 10 seconds for a path to appear in the device list. However this timeout is not enough or less than default device discovery timeout in most of the systems which cause certain device to be not accounted for the volume. This timeout has been lifted to 30seconds from this release and it should help to avoid mount issues due to device discovery. ([#78475](https://github.com/kubernetes/kubernetes/pull/78475), [@humblec](https://github.com/humblec))...

	if err != nil {
		return nil, toStorageErr(err)
	}
	return respBody, nil
}

// Stringer provides a canonicalized representation of network device.
func (client *storageRESTClient) String() string {
	return client.endpoint.String()
}

// IsOnline - returns whether client failed to connect or not.
func (client *storageRESTClient) IsOnline() bool {

        as arg with default value `None` to choose the I/O device for loading
        models and weights.
    *   Update `tf.saved_model.SaveOptions` with
        [`experimental_io_device`](https://www.tensorflow.org/versions/r2.3/api_docs/python/tf/saved_model/SaveOptions?hl=en)
        as arg with default value `None` to choose the I/O device for saving
        models and weights.

4. Instalar las dependencias de los paquetes en el archivo de requisitos.

    La opción `--no-cache-dir` le dice a `pip` que no guarde los paquetes descargados localmente, ya que eso solo sería si `pip` fuese a ejecutarse de nuevo para instalar los mismos paquetes, pero ese no es el caso al trabajar con contenedores.

    /// note | Nota

## Changelog since v1.3.7

### Other notable changes

* AWS: fix volume device assignment race condition ([#31090](https://github.com/kubernetes/kubernetes/pull/31090), [@justinsb](https://github.com/justinsb))



# v1.3.7

* Fix device mountable volume names in DSW to prevent races in device mountable plugin, e.g. local. ([#71509](https://github.com/kubernetes/kubernetes/pull/71509), [@cofyc](https://github.com/cofyc))