cog-sdxl
cog-sdxl copied to clipboard
replicate
Stable Diffusion XL training and inference as a cog model
I would like to suggest considering the support for some commonly used styles, as mentioned in the reference link: [https://huggingface.co/spaces/google/sdxl](https://huggingface.co/spaces/google/sdxl). This could potentially enhance the user experience. If it is...
[https://github.com/huggingface/hf_transfer](https://github.com/huggingface/hf_transfer) Just needs two lines of code 1) `- "hf_transfer"` to `cog.yaml` `python_packages` 2) `os.environ["HF_HUB_ENABLE_HF_TRANSFER"] = "1"` to the top of `train.py` and/or `predict.py`
I know that this template, as of now, doesn't support training multiple concepts. I would love to help out with a PR, but would need some guidance as to how...
On localhost, it is working fine, i am currently using free version of the api.
downloading url: weights downloading to: ./trained-model Error downloading file: Head "weights": unsupported protocol scheme "" Traceback (most recent call last): File "/root/.pyenv/versions/3.9.17/lib/python3.9/site-packages/cog/server/worker.py", line 185, in _setup run_setup(self._predictor) File "/root/.pyenv/versions/3.9.17/lib/python3.9/site-packages/cog/predictor.py", line...
When trying to run the SDXL preprocessor it can't find the model that's in the sample code (i.e. "replicate/sdxl_preprocess:bd1158a5052ed46176da900ad7e2a80ea04a3c46196d93f9e1db879fd1ce7f29"). The link to the model in the documentation is broken as...
When using token prefixes like `A photo of TOK, `, the BLIP caption outputs no longer received the substitution token uppercasing treatment because of the punctuation. Using re's word boundary...
Just a heads up that I have started work on a webui for cog training similar to the one for kohya_ss. If any of you guys are interested to contribute...
In `predict.py` feature extractor is loaded from a cache folder at line 154 `self.feature_extractor = CLIPImageProcessor.from_pretrained(FEATURE_EXTRACTOR)` But that cache folder is never downloaded. It should be added in /script/download_weigths.py
