CLI option to select model

[skorokithakis]

I have created a model with Textual Inversion, and I was wondering if there can be an easy CLI option to select the model to generate from?

[brycedrennan]

I haven’t played with textual inversion. I assume your model is a small file added to the initial large model. Is that right?

[skorokithakis]

I think no, it's a 2 GB ckpt file that I think replaces the original model. I'm trying to see if I can add the option myself, but my GPU has very little VRAM so I can either have the windowing system loaded or run the script, which makes development hard...

[skorokithakis]

Looks like simply changing ckpt_path in load_model_from_config to the path of my model does the trick.

[skorokithakis]

I don't know how you want to handle passing the option from the CLI into that function, though.

[brycedrennan]

Thanks for the idea. I just pushed out 2.3.0 which lets you specify the model via --model-weights-path

[skorokithakis]

Thank you!