>>42964574why are you so keen on recreating the model? I dont really remember the exact settings but I think i just chose 0.5 multiplier, weighted sum and safetensor format.
maybe I chose no interpolation, but I doubt it.
>>42965085I just put together the 3 relevant models I liked, mixed them together and took the (what I thought of as the) simplest path. If I knew someone tried to recreate it later, I wouldve documented it. but im not stubborn about anything, I just dont remember these details
