将pytorch中的三维tensor另存为csv文件_将tensor中值依次写到csv-优快云博客

#!/usr/bin/env python3

import torch.nn as nn
import torch.nn.functional
import numpy as np
i
#此案例来源于：https://pytorch.org/docs/master/nn.html#torch.nn.LSTM
rnn = nn.LSTM(input_size=10,hidden_size=20,num_layers=2)#输入向量维数10, 隐藏元维度20, 2个LSTM层串联(若不写则默认为1）
input = torch.randn(5,3,10)#输入（seq_len,batch , input_size） 序列长度为5 batch为3 输入维度为10
#print(input)
h0 = torch.randn(2,3,20)#h_0(num_layers * num_directions, batch, hidden_size)  num_layers = 2 ，batch=3 ，hidden_size = 20
c0 = torch.randn(2,3,20)#同上
output, (hn,cn) = rnn(input, (h0,c0))

#保存hn
a = hn.detach().numpy()
#三维numpy无法输出为csv,必须用reshape重塑
b = a.reshape(-1,20)
np.savetxt('hn.csv',b,delimiter=',')
print(hn)
print(b)

输出如下：

C:\Users\Penghj\Anaconda3\envs\pytorch-py36\python.exe D:/PycharmProjects/LSTMprocess/LSTM_test.py
tensor([[[-7.3971e-02, -1.2780e-01, -1.4474e-01,  1.8253e-01, -1.9796e-01,
           1.0784e-01, -2.3064e-01, -3.4146e-01,  1.8185e-01, -8.9133e-04,
           4.0730e-02, -1.2608e-01,  5.8794e-02, -7.0834e-02, -1.8286e-01,
          -1.0784e-02,  1.9347e-01, -1.5460e-01,  8.6555e-02, -1.6798e-02],
         [ 9.6833e-02, -5.8035e-02,  1.7099e-01,  2.2732e-01,  1.3198e-02,
           3.5267e-03,  4.5223e-02, -1.5988e-01, -9.0247e-02,  7.3956e-02,
           7.6391e-02, -4.5348e-02,  7.6712e-02,  2.2441e-03, -1.1218e-01,
           7.3470e-02,  1.0081e-01, -3.8649e-02,  1.3708e-01,  2.1249e-01],
         [ 5.5192e-02,  1.4865e-02,  1.6241e-02, -9.0010e-03, -1.9215e-01,
          -4.5265e-02,  1.0673e-01, -1.2473e-01, -1.9864e-04, -1.5129e-01,
          -8.4610e-02, -4.1374e-02,  1.1179e-01, -7.7280e-02, -9.2104e-04,
           4.8305e-02,  1.4308e-02, -6.9500e-03,  2.7160e-01,  1.0379e-01]],

        [[-1.4114e-01,  2.4604e-02, -3.0190e-02,  9.0903e-02,  1.7423e-02,
          -4.3247e-02, -1.7116e-01, -1.2302e-01, -1.6867e-01,  6.4440e-03,
          -1.3868e-01, -4.2504e-02, -8.0390e-02,  4.2705e-03,  3.0945e-02,
          -7.8335e-02,  6.4501e-02,  5.7008e-03, -5.9228e-02, -3.5933e-02],
         [-6.2894e-02,  6.8849e-02,  1.5310e-02,  3.9452e-02, -4.1810e-02,
          -1.6745e-01, -2.1690e-01, -1.1189e-02, -1.0120e-01, -5.4169e-02,
          -1.2677e-01, -8.9171e-02, -5.1950e-02, -8.6439e-02,  7.8760e-03,
          -1.0244e-01,  6.7068e-02,  8.7209e-02, -1.7350e-02, -1.0928e-02],
         [-1.4949e-01,  4.2771e-02, -2.9070e-02,  4.4891e-02, -4.9142e-02,
          -1.6704e-01, -1.4158e-01, -1.0561e-02, -1.2310e-01, -1.1926e-02,
          -1.1779e-01, -1.2357e-02, -7.4506e-03, -2.5705e-02,  1.0918e-02,
          -7.2432e-02,  7.4882e-02,  5.3149e-02, -7.8296e-03, -4.9892e-02]]],
       grad_fn=<StackBackward>)
[[-7.39708394e-02 -1.27802983e-01 -1.44737855e-01  1.82530552e-01
  -1.97959319e-01  1.07843086e-01 -2.30639011e-01 -3.41463804e-01
   1.81853518e-01 -8.91325413e-04  4.07299958e-02 -1.26077503e-01
   5.87935671e-02 -7.08335489e-02 -1.82858154e-01 -1.07838241e-02
   1.93470061e-01 -1.54596388e-01  8.65550786e-02 -1.67978760e-02]
 [ 9.68326107e-02 -5.80345728e-02  1.70993403e-01  2.27316663e-01
   1.31975403e-02  3.52669228e-03  4.52234782e-02 -1.59884900e-01
  -9.02474746e-02  7.39560351e-02  7.63912499e-02 -4.53478247e-02
   7.67116323e-02  2.24413490e-03 -1.12175845e-01  7.34703988e-02
   1.00805841e-01 -3.86485495e-02  1.37081578e-01  2.12489918e-01]
 [ 5.51915988e-02  1.48648266e-02  1.62407588e-02 -9.00098588e-03
  -1.92146197e-01 -4.52653170e-02  1.06727183e-01 -1.24732286e-01
  -1.98639274e-04 -1.51294112e-01 -8.46096203e-02 -4.13742140e-02
   1.11793645e-01 -7.72797391e-02 -9.21042752e-04  4.83049154e-02
   1.43076414e-02 -6.94996910e-03  2.71601051e-01  1.03792876e-01]
 [-1.41142428e-01  2.46040653e-02 -3.01896222e-02  9.09029543e-02
   1.74230635e-02 -4.32474427e-02 -1.71161547e-01 -1.23024724e-01
  -1.68667182e-01  6.44403743e-03 -1.38677344e-01 -4.25040163e-02
  -8.03901553e-02  4.27051587e-03  3.09446547e-02 -7.83353224e-02
   6.45011663e-02  5.70079219e-03 -5.92282042e-02 -3.59332785e-02]
 [-6.28941953e-02  6.88489452e-02  1.53099252e-02  3.94519903e-02
  -4.18104976e-02 -1.67447835e-01 -2.16900289e-01 -1.11890212e-02
  -1.01196185e-01 -5.41693345e-02 -1.26771182e-01 -8.91706571e-02
  -5.19499257e-02 -8.64390358e-02  7.87597708e-03 -1.02441035e-01
   6.70683980e-02  8.72091055e-02 -1.73503663e-02 -1.09276902e-02]
 [-1.49493888e-01  4.27707098e-02 -2.90701501e-02  4.48911339e-02
  -4.91419397e-02 -1.67038664e-01 -1.41577035e-01 -1.05611794e-02
  -1.23095192e-01 -1.19260075e-02 -1.17788106e-01 -1.23572880e-02
  -7.45061925e-03 -2.57053599e-02  1.09177465e-02 -7.24323615e-02
   7.48823658e-02  5.31486273e-02 -7.82960933e-03 -4.98918593e-02]]

Process finished with exit code 0