Hi all, version 0.0.81 is live! This release includes:

Desktop:

• Adds a new Experimental llama.cpp backend. This has several months of llama.cpp fixes in it which should increase the stability, quality, and speed of the LLM generation. I've tested in on Mac and Windows with a NVIDIA GPU, but not on Linux or Windows ARM. Would love to hear how it works for you and if you have any issues. You can enable it in the Settings page under `Backend Version -> Experimental`.
• Fixes the issue where long conversations would become incoherent - there are still issues with the model forgetting parts of the previous conversation, but now it won't get stuck like it used to.

Thank you!