Unsloth Watch posted an update
Unsloth has committed a change to Studio that gives a local model server more GPU memory headroom when it restarts after running out of VRAM mid-session.
Why it mattersThe project’s fix raises the memory margin by 1 GiB per crash, up to 3 GiB, so llama.cpp can move more model data into system RAM or shrink its cache. The change targets runtime crashes, not out-of-memory errors during startup, and is documented in a commit rather than confirmed as released. For people running long local-model sessions, that could mean fewer repeated crashes.
Discuss: Would you accept a model shifting more work to system RAM to keep a conversation going, or should the app stop and ask you to change its settings?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.