I was trying to convert a big NumPy array to a string by using np.array_str() or str() but it turns out that the output does not contain all the values of the array.
weight=np.random.rand(5000,500)
print(weight.shape)
print(len(np.array_str(weight)))
print(len(str(weight)))
>>(5000,500)
>>443
>>443
Which is way less than the actual count of digits.
How can I get the complete array as a string? N.B. I want to write the sting in a text file.
str(weight) is just the console representation of the string, it does not convert all values to string values, instead it generates a string that can be printed to the console, e.g.:
>>> str(weight)
'[[0.61063338 0.76216809 0.09732382 ... 0.74375515 0.09525192 0.43705637]\n [0.70583367 0.25657426 0.53279926 ... 0.72487029 0.36734946 0.67419836]\n [0.95006557 0.84218583 0.75602697 ... 0.51730571 0.11963987 0.84481707]\n ...\n [0.35017875 0.16172616 0.18016095 ... 0.74119508 0.09725982 0.93651959]\n [0.41584527 0.62796044 0.65794606 ... 0.12521724 0.62784847 0.57898542]\n [0.72758274 0.99773068 0.15162054 ... 0.42649996 0.631018 0.446582 ]]'
Try converting your array to a list and then to a string instead:
str(weight.tolist())
To generate a tab- and line-separated representation (if you accept this), you can do
with open("test.txt", "w") as fh: weight.tofile(fh, sep="\t")
Another way is to dump it as json, which is the same syntax as python code for a list:
from json import dump
with open("test.txt", "w") as fh: dump(weight.tolist(), fh)
I used json to make sure that nothing gets lost.
You can try:
conv_str = [[str(j) for j in i] for i in weight]
