Llama.cpp configuration
New Conversation and AI Task entity names will be prefixed with their server name for easy identification
Added sampling parameters under llama.cpp configuration options on Conversation and AI Task entities
Added a separate option to include prior model thinking content in the message history, disabled by default:
- Some models require this to be passed back, while some models explicitly request this not be provided
What's Changed
- Llamacpp sampling params by @skye-harris in #83
Full Changelog: 1.7.0...1.8.0