Closed kungkook closed 10 months ago
Hi, could you please confirm that the RemoteCILP checkpoint is properly downloaded, and the path_to_your_checkpoints
variable is correctly set, such that f"{path_to_your_checkpoints}/{model_name}.pt"
could successfully point to the checkpoint file?
Hi, could you please confirm that the RemoteCILP checkpoint is properly downloaded, and the
path_to_your_checkpoints
variable is correctly set, such thatf"{path_to_your_checkpoints}/{model_name}.pt"
could successfully point to the checkpoint file?
It's settled, thank you
`import torch, open_clip from PIL import Image from IPython.display import display
modelname = 'RN50' # 'RN50' or 'ViT-B-32' or 'ViT-L-14' model, , preprocess = open_clip.create_model_and_transforms(model_name) tokenizer = open_clip.get_tokenizer(model_name)
path_to_your_checkpoints = 'checkpoints'
ckpt = torch.load(f"{path_to_your_checkpoints}/{model_name}.pt", map_location="cpu") message = model.load_state_dict(ckpt) print(message) model = model.cuda().eval()
i run the code , and then:
n Module.load_state_dict(self, state_dict, strict) 1971 r"""Copies parameters and buffers from :attr:state_dict
into 1972 this module and its descendants. If :attr:strict
isTrue
, then 1973 the keys of :attr:state_dict
must exactly match the keys returned (...) 1991RuntimeError
. 1992 """ 1993 if not isinstance(state_dict, Mapping): -> 1994 raise TypeError("Expected state_dict to be dict-like, got {}.".format(type(state_dict))) 1996 missing_keys: List[str] = [] 1997 unexpected_keys: List[str] = []TypeError: Expected state_dict to be dict-like, got <class 'torch.jit._script.RecursiveScriptModule'>.`