rinongal

Closing due to lack of activity. Feel free to reopen if you still need help.

Can you tell me a bit more about what your setup is? Do you care at all about the size of your files? You could use our unfrozen_config and just...

Yep. This seems like 0.1 is indeed the sweet spot. You could try playing around that region (e.g. maybe 0.2 would be better), but here you've got the quickest convergence...

Another thing that might help: Change this line https://github.com/rinongal/textual_inversion/blob/0a950b482d2e8f215122805d4c5901bdb4a6947f/ldm/modules/embedding_manager.py#L59 To: `get_embedding_for_tkn = partial(get_embedding_for_clip_token,embedder.transformer.text_model.embeddings.token_embedding)` (e.g. add `.token_embedding` at the end of the second argument)

They are trying to overfit - so overshooting the distribution is not a concern. They want to get to a point where the likeness is best as soon as possible,...

You can absolutely do something like that, yes. I actually had something similar to begin with (which you can see in messy commented out code [here](https://github.com/rinongal/textual_inversion/blob/main/ldm/modules/encoders/modules_bak.py) because I forgot to...

I'm guessing this is resolved, so I'll close it for now. Let me know if you need more help.

That's a good question :) We use sklearn for the classification itself so that shouldn't be an issue. The GAN inversion network might be a bit too much without a...

Unfortunately it seems like the GAN inversion step does indeed require a GPU. If you have pre-inverted codes, it's possible to run inference (regression or sorting) on the CPU with...

Hi, [StyleSpace Analysis](https://github.com/betterze/StyleSpace) have an eye gaze direction which you may be able to use. It uses the S-space and not the W-space, and I believe their code is based...