Quanting the Stardust V2 model. According to the model card this is a slightly different tune which you can see in the Usecase section from the original model card below. Highly recommend giving it a read-over and determining if you still want to try it before downloading. Either way, I intend to give it a shot.
Both Mistral and ChatML should work though I had better results with ChatML:
ChatML Example:
1"""<|im_start|>user
2Hi there!<|im_end|>
3<|im_start|>assistant
4Nice to meet you!<|im_end|>
5<|im_start|>user
6Can I ask a question?<|im_end|>
7<|im_start|>assistant
8"""
This model was merged using the
DARE TIES merge method using
Sao10K/MN-12B-Lyra-v3 as a base.
Special thanks to the SillyTilly and myself for helping me find the energy to finish this.
