Skip to content

FreeToken Desktop v0.2.0-beta.17

Pre-release
Pre-release

Choose a tag to compare

View all tags
@FlashML-org FlashML-org released this 04 Sep 08:56
· 6 commits to main since this release
v0.2.0-beta.17
e405b5e

What's Changed

Features

  • Settings page links to the changelog.
  • Console restores each model's last applied cache config on load.

Fixes

  • Engine install now resolves the wheel pair from a per-platform manifest on the FreeToken-Web beta release, falling back to the GitHub API only when the manifest is missing. Fixes the 403 rate-limit failures on unauthenticated installs.
  • Switching models waits for the stopped serve's port to be released instead of showing an error banner.
  • Daemon port moved to 19000 with a bind test at startup; a legacy daemon left running is adopted instead of colliding.
  • Library shows an error instead of a stand-in VRAM figure when the local GPU cannot be read.
  • NVML now loads via libnvidia-ml.so.1, fixing GPU detection on Fedora and other distros without the unversioned symlink.
Assets 10
Loading