Skip to main content

Exporting Fine-Tuned Models

After completion of training, you can export your converged model and push it to the intended platform space as a brand-new repository. Below are the basic export path instructions for mainstream frameworks.

Scheme A: Exporting via LLaMA-Factory​

After completing a training cycle in the Web GUI, select the [Export] tab in the navigation list to access configurables:

  1. Select the Best Checkpoint Choose the most optimal target version to transform from the list of training snapshots.

    checkpoint

  2. Establish Export Path and Target ID

    • Export dir: Local temporary directory path for storing the result.
    • CSGHub ID: The name of the target repository carrying the new fruits of labor. The format must embody the namespace (Example: your-username/brand-new-model-name).
    checkpoint

  3. Execute Export Execute with the export interface button. The system then merges the underlying framework and finetuned layers and automatically pushes to the new repository.

Export from MS-Swift​

Select the training checkpoint using the current image's UI/instructions and merge LoRA when appropriate. Export to a local directory first and verify configuration, tokenizer and weight files.

If the image provides integrated CSGHub upload, confirm the actual endpoint, Access Token and target repository. An HF-compatible option does not by itself identify the correct CSGHub destination.

Otherwise use model upload to upload the export directory with an explicit endpoint and model type. Prefer a separate destination for experiment tracking rather than overwriting unverified results.

Verify the target commit and whether the output is a full model or an adapter requiring the base model. Reload or download it before stopping or deleting the instance.