stabilityai

/

stable-diffusion-x4-upscaler

like 665

Follow

Stability AI 5,985

Hi,

I'm a first time user, so I'm running through the simple demo (deepfloyd_if_free_tier_google_colab.ipynb). It fails in the last step (1.6) with the following message:

│ in <cell line: 1>:1 │
│ │
│ /usr/local/lib/python3.10/dist-packages/diffusers/pipelines/pipeline_utils.py:1063 in │
│ from_pretrained │
│ │
│ 1060 │ │ │ │ loaded_sub_model = passed_class_obj[name] │
│ 1061 │ │ │ else: │
│ 1062 │ │ │ │ # load sub model │
│ ❱ 1063 │ │ │ │ loaded_sub_model = load_sub_model( │
│ 1064 │ │ │ │ │ library_name=library_name, │
│ 1065 │ │ │ │ │ class_name=class_name, │
│ 1066 │ │ │ │ │ importable_classes=importable_classes, │
│ │
│ /usr/local/lib/python3.10/dist-packages/diffusers/pipelines/pipeline_utils.py:451 in │
│ load_sub_model │
│ │
│ 448 │ │
│ 449 │ # check if the module is in a subdirectory │
│ 450 │ if os.path.isdir(os.path.join(cached_folder, name)): │
│ ❱ 451 │ │ loaded_sub_model = load_method(os.path.join(cached_folder, name), **loading_kwar │
│ 452 │ else: │
│ 453 │ │ # else load from the root directory │
│ 454 │ │ loaded_sub_model = load_method(cached_folder, **loading_kwargs) │
│ │
│ /usr/local/lib/python3.10/dist-packages/diffusers/models/modeling_utils.py:649 in │
│ from_pretrained │
│ │
│ 646 │ │ │ │ else: # else let accelerate handle loading and dispatching. │
│ 647 │ │ │ │ │ # Load weights and dispatch according to the device_map │
│ 648 │ │ │ │ │ # by default the device_map is None and the weights are loaded on th │
│ ❱ 649 │ │ │ │ │ accelerate.load_checkpoint_and_dispatch( │
│ 650 │ │ │ │ │ │ model, │
│ 651 │ │ │ │ │ │ model_file, │
│ 652 │ │ │ │ │ │ device_map, │
│ │
│ /usr/local/lib/python3.10/dist-packages/accelerate/big_modeling.py:486 in │
│ load_checkpoint_and_dispatch │
│ │
│ 483 │ │ ) │
│ 484 │ if offload_state_dict is None and device_map is not None and "disk" in device_map.va │
│ 485 │ │ offload_state_dict = True │
│ ❱ 486 │ load_checkpoint_in_model( │
│ 487 │ │ model, │
│ 488 │ │ checkpoint, │
│ 489 │ │ device_map=device_map, │
│ │
│ /usr/local/lib/python3.10/dist-packages/accelerate/utils/modeling.py:1116 in │
│ load_checkpoint_in_model │
│ │
│ 1113 │ │ │ │ │ set_module_tensor_to_device(model, param_name, "meta") │
│ 1114 │ │ │ │ │ offload_weight(param, param_name, state_dict_folder, index=state_dic │
│ 1115 │ │ │ │ else: │
│ ❱ 1116 │ │ │ │ │ set_module_tensor_to_device(model, param_name, param_device, value=p │
│ 1117 │ │ │
│ 1118 │ │ # Force Python to clean up. │
│ 1119 │ │ del checkpoint │
│ │
│ /usr/local/lib/python3.10/dist-packages/accelerate/utils/modeling.py:142 in │
│ set_module_tensor_to_device │
│ │
│ 139 │ if "." in tensor_name: │
│ 140 │ │ splits = tensor_name.split(".") │
│ 141 │ │ for split in splits[:-1]: │
│ ❱ 142 │ │ │ new_module = getattr(module, split) │
│ 143 │ │ │ if new_module is None: │
│ 144 │ │ │ │ raise ValueError(f"{module} has no attribute {split}.") │
│ 145 │ │ │ module = new_module │
│ │
│ /usr/local/lib/python3.10/dist-packages/torch/nn/modules/module.py:1614 in getattr │
│ │
│ 1611 │ │ │ modules = self.dict['_modules'] │
│ 1612 │ │ │ if name in modules: │
│ 1613 │ │ │ │ return modules[name] │
│ ❱ 1614 │ │ raise AttributeError("'{}' object has no attribute '{}'".format( │
│ 1615 │ │ │ type(self).name, name)) │
│ 1616 │ │
│ 1617 │ def setattr(self, name: str, value: Union[Tensor, 'Module']) -> None: │
╰──────────────────────────────────────────────────────────────────────────────────────────────────╯
AttributeError: 'Attention' object has no attribute 'key'

It looks like library errors. Please help, thanks!