Passing job variables to ACI push work pool in a nested flow

I would like to override the number of CPUs in a nested flow. Here is my inner flow.

import os

from prefect import flow, get_run_logger, task
from prefect.deployments import DeploymentImage
from prefect.task_runners import SequentialTaskRunner
import time

# Define the flow that actually runs the computations
@flow(task_runner=SequentialTaskRunner())
def log_hello(name):
    logger = get_run_logger()
    logger.info(f"{name} said hello!")
    time.sleep(30)
    logger.info("Done!")
    return name


if __name__ == "__main__":
    log_hello.deploy(
        name="aci",
        work_pool_name="my-aci-pool5",
        image=DeploymentImage(name="my-image", platform="linux/amd64"),
    )

And here is the outer flow trying to set job variables for the inner flow:

from prefect import flow, task
from prefect.deployments import DeploymentImage, run_deployment


# Helper task to run the deployment of log_hello
@task
def log_hello_run_deployment(name):
    run_deployment(name="log-hello/aci", parameters={"name": name},
                   job_variables={"cpu": 3})


# The upstream flow that calls log_hello
@flow
def log_hello_flow():
    names = ["Marvin", "Alice", "Bob"]
    log_hello_run_deployment.map(names)


if __name__ == "__main__":
    deployment = log_hello_flow.deploy(
        name="distributed",
        work_pool_name="my-aci-pool5",
        image=DeploymentImage(name="my-image", platform="linux/amd64"),
    )
    run_deployment('log-hello-flow/distributed')

The instances that are created have 1 CPU, as stated in the default job template I use. I can change the number of CPUs in the UI, but this would affect all instances in the pool. Am I doing something wrong? How can I change the number of CPUs only for the inner flow?