Describe the bug The wandb.watch() causes the exception below when we want to save the model using torch.save(). There exists workaround (passing to torch.save() an argument: pickle_module=dill) but when log_graph=True is used in the wan...
我用torch.save会报错,直接DataSource.write_pickle(model),会提示<trackeback: AttributeError: Can't pickle local object 'get_cosine_schedule_with_warmup.<locals>.lr_lambda'>解答已经解决了,model.schedulers = None就可以保存了标签AttributeErrorTransformer 评论 已经解决了,model.schedulers = None就可以...
AttributeError: Can't pickle local object 'DataLoader.init..' Anyone had similar issues ? read pickle cant pickle lambda functions but I wonder why the default implementation of pytorch is made this way ? Environment OS: Windows Python version: 3.6 PyTorch version: 1.1 CUDA/cuDNN version: 10...
AttributeError: Can't pickle local object 'DataLoader.init..' Anyone had similar issues ? read pickle cant pickle lambda functions but I wonder why the default implementation of pytorch is made this way ? Environment OS: Windows Python version: 3.6 PyTorch version: 1.1 CUDA/cuDNN version: 10...
in save _save(obj, opened_zipfile, pickle_module, pickle_protocol, _disable_byteorder_record) File "/home/dujiang/miniconda3/envs/yolo/lib/python3.10/site-packages/torch/serialization.py", line 864, in _save pickler.dump(obj) AttributeError: Can't pickle local object 'classify_transforms....