ValueError: shapes (100,) and (1,100) not aligned: 100 (dim 0) != 1 (dim 0)

ValueError: shapes (100,) and (1,100) not aligned: 100 (dim 0) != 1 (dim 0)¶

• dot, cosine_similarity 등 매트리스끼리 연산할때, 두 매트리스의 shape가 연산규칙에 맞지 않다면 ValueError 가 발생하네요.

In [1]:

import numpy as np

• (100) X (1,100) 어레이는 에러가 산출되네요

In [2]:

a = np.random.rand(100)  
b = np.random.rand(1, 100) 
np.dot(a, b)

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
Cell In[2], line 3
      1 a = np.random.rand(100)  
      2 b = np.random.rand(1, 100) 
----> 3 np.dot(a, b)

ValueError: shapes (100,) and (1,100) not aligned: 100 (dim 0) != 1 (dim 0)

• (1, 100) X (1, 100) 어레이도 에러를 반환하네요

In [3]:

a = np.random.rand(1, 100)  
b = np.random.rand(1, 100)  
np.dot(a, b)

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
Cell In[3], line 3
      1 a = np.random.rand(1, 100)  
      2 b = np.random.rand(1, 100)  
----> 3 np.dot(a, b)

ValueError: shapes (1,100) and (1,100) not aligned: 100 (dim 1) != 1 (dim 0)

• (100, 1) X (1,100) 어레이는 계산이 됩니다.

In [4]:

# Example arrays
a = np.random.rand(100, 1)  # This will create an array of shape (100,)
b = np.random.rand(1, 100)  # This will create an array of shape (1, 100)
np.dot(a, b)

Out[4]:

array([[0.25148646, 0.25288578, 0.02490737, ..., 0.12144668, 0.31535732,
        0.28806871],
       [0.13723606, 0.13799967, 0.01359194, ..., 0.06627341, 0.17209036,
        0.15719899],
       [0.02842853, 0.02858672, 0.00281558, ..., 0.01372858, 0.03564862,
        0.03256387],
       ...,
       [0.60409463, 0.60745594, 0.0598299 , ..., 0.2917266 , 0.75751857,
        0.69196873],
       [0.38783055, 0.38998852, 0.03841097, ..., 0.18728934, 0.48632917,
        0.44424598],
       [0.21079122, 0.21196411, 0.02087689, ..., 0.10179432, 0.26432657,
        0.24145378]])

• 구글링해보니 어레이 선언은 일괄적으로 하고 필요한 어레이에 transpose 해 계산해 주는게 대세인 것 같습니다.

In [5]:

a = np.random.rand(1, 100)  
b = np.random.rand(1, 100) 
np.dot(a.T, b)

Out[5]:

array([[0.10471007, 0.07425017, 0.20002182, ..., 0.14549293, 0.39086824,
        0.41040463],
       [0.02366827, 0.01678323, 0.04521217, ..., 0.03288667, 0.08835037,
        0.09276631],
       [0.12110494, 0.08587581, 0.23134003, ..., 0.16827333, 0.45206802,
        0.47466331],
       ...,
       [0.03313391, 0.02349533, 0.06329386, ..., 0.04603902, 0.1236843 ,
        0.1298663 ],
       [0.07990834, 0.0566632 , 0.15264445, ..., 0.11103132, 0.29828679,
        0.31319577],
       [0.19737529, 0.1399593 , 0.37703503, ..., 0.27424972, 0.7367747 ,
        0.77360021]])