引導式測驗 · 中等 等級
在 Hugging Face 空間測驗中部署組合演示
使用正確的 Space SDK、可重現的依賴關係、受保護的秘密、明確的限制和即時檢查來規劃可共享的模型演示。
問題 1 的 8
Which Space SDK lets you define system packages and startup commands in a Dockerfile?
預覽每個問題8 問題
當您參加測驗時,就會出現答案和解釋。
- Which Space SDK lets you define system packages and startup commands in a Dockerfile?
- Gradio SDK conventions
- Static HTML SDK
- Setting suggested_hardware to a GPU
- Docker SDK
- When is a static HTML Space a natural fit?
- An app requiring a Python server process and installed system packages
- A client-side page that does not need server-side model inference
- A model that must run locally on a Space GPU for each request
- A custom Linux inference runtime inside the Space
- For a server-side Gradio or Docker Space, where should a private API key be stored and accessed?
- In a committed source file
- In Space Secrets and read by server-side code
- In the browser JavaScript bundle
- As a public Space variable
- Which check provides evidence that the deployed Space works end to end?
- Run the deployed URL with representative valid, invalid and boundary inputs
- Inspect the build log without opening the app
- Check repository metadata and commit status only
- Test only a local development copy
- Which information most directly helps users interpret a model result?
- The expected input and output plus known limitations
- A claim that the model cannot fail
- A guarantee that a successful build proves accuracy
- A statement that the interface removes licensing concerns
- Why pin dependencies and test a clean build?
- To guarantee identical predictions on all hardware
- To reduce version drift and catch missing setup assumptions
- To make model outputs independent of data
- To replace evaluation of the live app
- Which detail should the demo disclose about uploaded user content?
- Whether inputs are logged or retained
- The GPU model used by every other Space
- The editor used to write the interface
- The number of lines in the README
- Which statement about a hosted portfolio demo is most accurate?
- A hosted demo is not automatically a production-ready service
- A successful build proves reliability under production load
- A public URL guarantees model validity
- A Gradio interface resolves model licensing concerns