diffusion-models-class

diffusion-models-class copied to clipboard

In this notebook, originally there is a line of code: image_pipe.save_pretrained("my_pipeline") , which will generate a saved model as 'diffusion_pytorch_model.safetensors'. However, it will cause errors in the following line pipeline = DDPMPipeline.from_pretrained(hub_model_id)

After investigation, we have to specify one more parameter when saving model as follows, image_pipe.save_pretrained("my_pipeline", safe_serialization=False)

Then, it should be work when loading our own trained model.