UnpicklingError while loading the model with PyTorch

[The-Gupta]

`import torch
import torch.autograd
import torch.nn
import torch.multiprocessing
import torch.utils
import torch.legacy.nn
import torch.legacy.optim

xp = torch.load(r"D:\SDS\1_MachineLearning\amr-eager-master\LDC2015E86\reentrancies.dat")`

Traceback (most recent call last):

File "", line 9, in
xp = torch.load(r"D:\SDS\1_MachineLearning\amr-eager-master\LDC2015E86\reentrancies.dat")

File "D:\Anaconda3\envs\amr-eager\lib\site-packages\torch\serialization.py", line 261, in load
return _load(f, map_location, pickle_module)

File "D:\Anaconda3\envs\amr-eager\lib\site-packages\torch\serialization.py", line 399, in _load
magic_number = pickle_module.load(f)

UnpicklingError: invalid load key, '�'.

reentrancies.dat model weights could be downloaded from here.
What should I change?

[The-Gupta]

@tastyminerals, could you help me with this?

[tastyminerals]

torch.load works with torch models, they have .t7 extension. You are loading a .dat file. This does not look like a torch model.

@mdtux89 is correct.
Yep, only now noticed, you are using python. This is a Torch repo not PyTorch. These are two separate frameworks, Torch uses Lua, PyTorch uses Python. You need to convert this model to PyTorch first. Please read this: https://discuss.pytorch.org/t/convert-import-torch-model-to-pytorch/37/2
and this: https://github.com/clcarwin/convert_torch_to_pytorch

[mdtux89]

It is a Lua Torch model, not pytorch

[The-Gupta]

require 'dp'
require 'optim'
require 'nngraph'
xp = torch.load(r"D:/SDS/1_MachineLearning/amr-eager-master/LDC2015E86/labels.dat")

Loads with LuaJIT on Linux.

@mdtux89, okay, now I know I can't use this model directly in Python. I had visited the links suggested by @tastyminerals, but could not understand how to convert the model. Could you convert or suggest how to?

[mdtux89]

https://discuss.pytorch.org/t/convert-import-torch-model-to-pytorch/37/2 I am out of office and as I said before via email I am not able to provide you with support on this at the moment