설정 - CoreWeave Docs

The examples on this page omit auth. To authenticate with your W&B API key, install cwsandbox[wandb] and pass auth=AuthStrategy.WANDB to Sandbox.run(), Sandbox(), or Sandbox.session(). See Get started.

This tutorial shows you how to configure a sandbox at startup by passing options to Sandbox.run(). It’s for developers who have already created a basic sandbox and want to control the container image, resource allocation, and pre-populated files for their workload.

Pass configuration to `Sandbox.run()`

Configuration options control how the sandbox is provisioned. Pass them as keyword arguments directly to Sandbox.run():

from cwsandbox import Sandbox

with Sandbox.run(
    container_image="python:3.11",
    max_lifetime_seconds=300,
    tags=["my-app"],
) as sandbox:
    sandbox.exec(["python", "--version"]).result()

Resources

Configure CPU and memory to match your workload’s needs and avoid over-provisioning or under-provisioning. Use ResourceOptions to set requests and limits:

from cwsandbox import ResourceOptions, Sandbox

with Sandbox.run(
    resources=ResourceOptions(
        requests={"cpu": "2", "memory": "2Gi"},
        limits={"cpu": "2", "memory": "2Gi"},
    ),
) as sandbox:
    sandbox.exec(["python", "compute.py"]).result()

ResourceOptions uses Kubernetes resource syntax:

Resource	Format	Examples
CPU	Cores or millicores	`"1"`, `"2"`, `"500m"` (0.5 CPU)
Memory	Bytes with unit suffix	`"512Mi"`, `"1Gi"`, `"2Gi"`

Mounted files

Mounted files let you ship configuration, scripts, or other static content into the sandbox before it starts, so your workload has what it needs the moment it runs. Pre-populate files at sandbox startup:

with Sandbox.run(
    mounted_files=[
        {"mount_path": "/app/config.json", "file_content": b'{"debug": true}'},
        {"mount_path": "/app/script.py", "file_content": b"print('hello')"},
    ]
) as sandbox:
    sandbox.exec(["python", "/app/script.py"]).result()

Each entry takes mount_path, the absolute path inside the sandbox, and file_content, the file’s contents as bytes. A str is encoded as UTF-8 for you. Mounted files are read-only. Use write_file() for files that need modification.

Next steps

You now have a sandbox that starts with the container image, resource allocation, and pre-mounted files your workload requires. For reusable configuration across multiple sandboxes, use SandboxDefaults. See the sandbox configuration guide for all ResourceOptions fields, GPU resources, QoS classes, and config-library interop.

​Pass configuration to Sandbox.run()

​Resources

​Mounted files

​Next steps

Pass configuration to `Sandbox.run()`

Resources

Mounted files

Next steps