Skip to content

Show model sampling defaults and allow per-request overrides in the Windows client #547

Description

@dai-ych

Before you start

  • I have read the Roadmap and this is not already planned there.
  • I have read the FAQ.
  • I have searched existing issues and found no duplicate.

Applies to

Desktop app

What problem does this solve

When a request does not include sampling parameters, generation uses defaults configured when the model server starts. The Windows client does not show those defaults, so users cannot tell which values are active or adjust them from the client. This makes it harder to understand or control generation behavior compared with clients such as LM Studio.

Proposed solution

Add a sampling section to the Windows client that displays the active model/server defaults for Temperature, Top P, Top K, Min P, Repetition Penalty, and Presence Penalty. Allow the calling client to override these values for a request, and indicate whether each effective value comes from the model/server default or a request-level override. If a request omits a parameter, use the configured model/server default.

Alternatives considered

Keep sampling defaults hidden in the model startup configuration and require users to inspect or change server settings outside the Windows client. This does not make the active values visible to users of the client.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    Desktopproblem related to FreeToken DesktopfeatureNew feature or request

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions