Amitg1

same here. any suggestion on how to debug it?

Any news regarding this one? how can we unload LoRa with only one encoder? Thanks!

Yes, it crashes. @strint diffusers 0.29.2 onediff 1.2.0 onediffx 1.2.0 torch 2.4.0 changed to ``` #load_and_fuse_lora(pipe, "nerijs/pixel-art-xl", lora_scale=1) pipe.load_lora_weights("nerijs/pixel-art-xl") res = pipe(**input_request).images[0] # unfuse_lora(pipe) pipe.unload_lora_weights() ``` error --------------------------------------------------------------------------- RuntimeError Traceback...

Hi, we are getting this error only from diffusers > 0.31. when diffusers == 0.31 we can load LoRa and perform inference. this setup works: diffusers 0.31.0 transformers 4.48.1 torch...

not the same person :) our example is pretty much the same, except we are not executing quantize and freeze and are using the following class to load the quantized...

Any news here?