Benjamin Bossan

I did some more digging and managed to make the unexpected keys disappear. For this, I had to use slightly different names for the arguments: ```diff - te2_state_dict[diffusers_name] = state_dict.pop(key)...

> Hmm. It's weird that we need to do this for this particular one. I will see if I can come up with something better. Yes. IIUC, one is using...

Then my best bet would be that some entries need to be added to this dict: https://github.com/huggingface/diffusers/blob/edf5ba6a17d012411c1fe3ceaf24f71f1899bc48/src/diffusers/utils/state_dict_utils.py#L67

Sorry for the late reply. @makrai I think the blog post contains an error in this line: `loss_fct = torch.nn.CrossEntropyLoss(weight=torch.tensor([neg_weights, pos_weights], device=model.device, dtype=logits.dtype))` When having multiple GPUs available, transformers `Trainer`...

Could you please share the full script and the command to run it?

> Still have another question; I notice that `ranknum` does not correspond to the dimension of `lora_E`, do you know why? No, not sure. Again, if you can share code,...

This should not happen. Could you please share some code to reproduce this error? If you're on the latest PEFT version, you can also run `model.get_model_status()` and `model.get_layer_status()` to help...

Hey, could you please paste the code as text, otherwise I'd have to copy everything by hand if I want to reproduce :) Also, if you call `base_model.add_adapter(...)`, you're adding...

Okay, thanks for trying. Since you use some private adapters, I can't really reproduce, unless you can share your adapters. One thing to try out would be to use PEFT...

> Regardless of my private adapters is this approach working with you any adapter you can access? Yes, it's working, here is a simple test: ```python import torch from peft...