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.
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.