Does anyone know how to weighted merge two Flux models?

[araleza]

Back when I was fine tuning SDXL, I used to train the model, and then do a (e.g.) 60%-40% merge back with SDXL base, then train again. I found it helped image quality a lot, as the junk that built up during the first training run was cleared away, with the second training run seeming to be better able to reinforce the actual 'best way' to add the trained concept.

I used to use automatic1111-webui for this. But that doesn't work with Flux models, and I don' t know of an alternative that does. Does anyone know?