Where is Ambolio lineage source?
The word Ambolio lineage (v1.0.6 / v1.0.8) literally does not exist on the internet 😃
I thought it's some repo on github but it was not there. Also nothing inside here as well.
Could you please shed some light on this.
Good catch — the name really is confusing when the original repo is unfindable.
Ambolio/ninfer-4090-windows was the original Windows/Ada port line of the engine
whose PQ2_0_G128 / PTQ1_0_G128 dialect this artifact uses. The author has since
removed all of his public repos (the GitHub account shows public_repos = 0, and the
HF mirror 404s too). So "search Ambolio" finding nothing is expected — the lineage
now only survives through forks.
I've just put the full conversion toolchain + a "where to get the engine" guide up on
GitHub: https://github.com/5258MF/ninfer-swift-bonsai2-converter
engine/README.md— why Ambolio 404s, and the verified forks you can build fromconverter/+REPRODUCE.md— the packer and step-by-step repro, if you want to
re-create the artifacts from the GGUF yourself- READMEs in this HF repo now point there too (see Engine compatibility section)
I have a question regarding my RTX 4070 (12GB VRAM): which models could I convert into LoRAs using that code?
My code is used to convert the bonsai2 model from GGUF to ninfer format, and it has nothing to do with LoRAs. It only involves model format conversion. For your configuration, bonsai2 is a good choice.