/Python/Pickle.load from gpu device to cpu

Pickle.load from gpu device to cpu

#if you are using pickle load on a cpu device, from a gpu device
#you will need to override the pickle load functionality like this:
#note: if loading directly in torch, just add map_location='cpu' in load()

class CPU_Unpickler(pickle.Unpickler):
    def find_class(self, module, name):
        if module == 'torch.storage' and name == '_load_from_bytes':
            return lambda b: torch.load(io.BytesIO(b), map_location='cpu')
        else: return super().find_class(module, name)

#contents = pickle.load(f) becomes...
contents = CPU_Unpickler(f).load()

Source: github.com

Similar codes

1 vote254 views1 answer
Max pooling tf keras
Python
1 vote220 views1 answer
Dataframe from another dataframe
Python
0 vote204 views1 answer
How to change scenes in python
Python
0 vote179 views1 answer
Close all tables python
Python
0 vote210 views1 answer
Blender python set object location
Python
0 vote224 views2 answers
Python aggregate count and sum
Python
0 vote206 views1 answer
Django queryset to form
Python
0 vote228 views1 answer
Générer des valeurs de 0 à n python liste
Python
2 votes202 views2 answers
Pandas replace null values with values from another column
Python
0 vote207 views1 answer
Convert numpy array to tfrecord and back
Python