Been trying to figure this thing out and I just cant seem to find heads or tails in all of this.
I’ve gotten far enough to figure out that lfs splits large models into smaller ones and names them “[pytorch_model-00001-of-00007.bin]” and so on. However, I am really not sure how to merge them back together into a single model.
Been trying to clone with the git lfs thing, however once they were cloned, they stayed in multiple shards. So I assume I’m supposed to do something more? (Or perhaps it didnt assemble the shards due to an error on my end?)
I tried looking through the help section, but that simply confused me more. I’ve been using Huggingface to find interesting a.i models so far, using it as a sort of “Git, but with better indexing”. But, after reading the how-to i’ve come to the conclusion that thats not what this site is about.
Am I supposed to download Accelerate to make use of sharded models?
I am a bit lost which might be evident from my post