Added SAG - Self Attention Guidance option to model selection. From my experiments this slows generation down by ~15%, but improves quality significantly, so worth it
Added FreeUv2
Added LLM system prompt - Can now tweak the system prompt that is sent with an LLM request in the LLM widget
Added 2 new Mistral models to model list - Mixtral 8x7b and Mistral Large
minor fix: Change the IPAdapter FaceID lora loader to what I think is the default Lora loader node rather than a custom node