Skip to main content

Command Line Upload and Download

Prerequisites​

  • A valid MatrixHub account.
  • Joined the target project with read/write permissions for the model repository (Admin or Developer permissions).
  • Hugging Face CLI installed locally (hf command available).
  • Network access to the MatrixHub service endpoint.

Uploading Models​

  1. Log in to the platform, go to Project Management, and select the target project.

  2. Open the Model Repository tab and click Create Model.

    Create Model

  3. Fill in the model name, confirm creation, and enter the model details page.

    After Create Model

  4. Configure the service endpoint in your local terminal.

    export HF_ENDPOINT="https://<your-matrixhub-endpoint>"

  5. Use hf upload to upload the local model directory.

    hf upload <project-name>/<model-name> ./<local-model-dir>

  6. Return to the model details page and refresh to confirm the uploaded files appear in the list.

    After Upload

note
  • If the model name is already taken, please choose a different name and try again.
  • Uploading large models for the first time may take a while; please wait for the command to complete.

Downloading Models​

  1. Enter the target model details page and click Download Model.

  2. Copy the download command from the popup and execute it in your local terminal.

    export HF_ENDPOINT="https://<your-matrixhub-endpoint>"
    hf download <project-name>/<model-name>

  3. After the command completes, the terminal will output the download directory path.

  4. Open the local download directory and verify that the model files are complete and usable.

Proxy Project Upload and Download​

  1. Create a proxy project first (e.g., proxy-demo).

  2. After creating a model in the proxy project, you can perform uploads and downloads via the command line against the proxy site.

    export HF_ENDPOINT="https://<your-matrixhub-endpoint>"

  3. Download a model from the proxy project (example).

    hf download proxy-project/Qwen3-ASR-0.6B

  4. Download a model from a proxy project mapped to a Hugging Face organization (e.g., prajjwal1) (example).

    hf download prajjwal1/bert-tiny

  5. Upload a local model to the proxy project (example).

    hf upload proxy-project/tiny-model ./<local-model-dir>
note
  • Once the proxy project is created, you can access proxy site model repositories using hf download and hf upload.
  • hf upload requires both the remote repository and the local path: hf upload <project>/<model> <local_path>.

Model Files​

Model Files Interface

Downloading Single Model Files​

  1. Enter the model details page and switch to the Model Files tab.
  2. Click Download on the target file's row.
  3. After the browser completes the download, open the file to verify the content.

File Search and Browsing​

  1. Use the search box in the Model Files page to enter keywords (e.g., .git, tokenizer).
  2. Observe the filtered results to ensure the returned files match expectations.
  3. If there are many files, click Load More to view the full list.

Branch/Version Switching​

  1. Enter the Model Files page of the model details.
  2. Select the target branch (e.g., main, test) in the branch selector.
  3. Verify that the file list after switching matches the content of that branch.