coo_matrix TypeError: data type not understood

Question

Steps

import torch
import numpy as np
from scipy.sparse import coo_matrix

row = torch.ones([10000], dtype=torch.float32)
col = torch.ones([10000], dtype=torch.float32)
data = torch.ones([10000], dtype=torch.float32)
fn = 5120
bm = 10000

coo_matrix((data, (row, col)), shape=(fn, bm))

Error Produced

Traceback (most recent call last):
File "", line 1, in
File "/usr/local/lib64/python3.8/site-packages/scipy/sparse/coo.py", line 161, in init
self.data = np.array(obj, copy=copy, dtype=data_dtype)
TypeError: data type not understood

Versions
Python 3.8.8
numpy 1.17.5
torch 1.7.1

Answer 1

There are two ways to solve this problem:-

1. Use a tensor based function that accepts the tensors as default (Use torch.sparse_coo_tensor )
   
2. Convert the tensors to numpy arrays using

tensor_data.cpu().detach().numpy()