fast.ai - manual prediction

fastai-manual_input.py

# single prediction on last classifier
learn.load('clas_2')
m = learn.model
#set batch size to 1
m[0].bs=1
#turn off dropout
m.eval()
#reset hidden state
m.reset()  

#sentence to be classified
sen = "guten morgen schon novemberlicht da mag man gar nicht glauben was gestern wetter war"
idxs = np.array([[stoi[p] for p in sen.strip().split(" ")]])

#converting sentence into numerical representation
print('numeric representation: ' + str(idxs))
idxs = np.transpose(idxs)

print('Value of idxs: ' + str(idxs)) 

#get predictions from model
p = m(VV(idxs))
print('Prediction: ' + str(p))
print(sen + ' - classification: ' + str(to_np(torch.topk(p[0],1)[1])[0]))

prob = F.softmax(p[0])
print('Probability rating for sentence: ' + str(prob))

fastai-manual_output

numeric representation: [[ 1611  2295   179     0    97  5670    96  1668    47  2236   180 28919  5816    33]]
Value of idxs: [[ 1611]
 [ 2295]
 [  179]
 [    0]
 [   97]
 [ 5670]
 [   96]
 [ 1668]
 [   47]
 [ 2236]
 [  180]
 [28919]
 [ 5816]
 [   33]]
Prediction: (Variable containing:
 0.2814 -0.0482 -0.0967
[torch.FloatTensor of size 1x3]
, [Variable containing:
( 0  ,.,.) = 
  3.2617e-04  1.3194e-03  3.9376e-06  ...   7.5627e-04  2.5742e-04  4.6105e-04

( 1  ,.,.) = 
  5.3593e-04 -2.3461e-04 -7.1204e-05  ...   6.6057e-05  1.1183e-06 -7.9078e-05

( 2  ,.,.) = 
  2.8301e-07 -5.0946e-04  2.8877e-05  ...   4.4297e-04 -1.3625e-05 -6.2073e-07
 ... 

( 11 ,.,.) = 
 -4.2971e-09  1.4740e-01 -2.1405e-01  ...  -1.3160e-01 -3.5512e-04  1.3877e-01

( 12 ,.,.) = 
 -7.4684e-04 -4.1670e-01 -6.1688e-06  ...  -1.2905e-04 -2.9628e-05  3.0681e-02

( 13 ,.,.) = 
 -1.8500e-05 -5.9534e-04 -2.3576e-07  ...  -1.4994e-06 -2.1938e-05 -2.2080e-04
[torch.FloatTensor of size 14x1x1150]
, Variable containing:
( 0  ,.,.) = 
 -7.5986e-03  8.6798e-04 -6.7351e-03  ...   6.2188e-04  3.4969e-02 -9.1077e-02

( 1  ,.,.) = 
 -2.3921e-03  2.4249e-03 -1.7347e-03  ...   1.0938e-02 -3.6272e-04 -1.5099e-02

( 2  ,.,.) = 
 -2.7945e-02  9.5424e-04 -2.8550e-03  ...   9.9346e-05 -4.6923e-03 -1.5932e-02
 ... 

( 11 ,.,.) = 
  7.7081e-05  1.1043e-04  3.5880e-03  ...   3.4046e-03 -3.2344e-01 -7.6625e-04

( 12 ,.,.) = 
  3.6866e-03  3.5075e-03  7.5826e-04  ...  -2.6639e-02 -2.2028e-01 -2.5222e-03

( 13 ,.,.) = 
 -1.3276e-02  3.8389e-02  1.6594e-04  ...  -2.6982e-03 -1.2232e-01 -6.5385e-03
[torch.FloatTensor of size 14x1x1150]
, Variable containing:
( 0 ,.,.) = 
 -3.4815e-02  1.2264e-03  2.1081e-01  ...   1.3807e-01 -1.8228e-01 -4.8058e-03

( 1 ,.,.) = 
 -3.8202e-02 -1.0102e-01  1.4506e-01  ...   7.9288e-03 -1.2543e-01 -4.7196e-03

( 2 ,.,.) = 
 -2.1076e-02 -5.5695e-02  1.0110e-01  ...   9.5715e-03 -9.4924e-02 -1.6425e-03
... 

(11 ,.,.) = 
 -1.6449e-02 -8.2472e-04  3.1275e-02  ...   2.5666e-02 -9.2902e-02 -1.3512e-01

(12 ,.,.) = 
 -3.0238e-02 -2.3225e-03  1.2036e-01  ...   1.6667e-02  1.6427e-01 -2.0484e-02

(13 ,.,.) = 
 -2.2449e-01 -2.6965e-03  3.8396e-02  ...   1.3158e-03  4.0717e-02 -1.3653e-02
[torch.FloatTensor of size 14x1x400]
], [Variable containing:
( 0  ,.,.) = 
  3.2617e-04  1.3194e-03  3.9376e-06  ...   7.5627e-04  2.5742e-04  4.6105e-04

( 1  ,.,.) = 
  5.3593e-04 -2.3461e-04 -7.1204e-05  ...   6.6057e-05  1.1183e-06 -7.9078e-05

( 2  ,.,.) = 
  2.8301e-07 -5.0946e-04  2.8877e-05  ...   4.4297e-04 -1.3625e-05 -6.2073e-07
 ... 

( 11 ,.,.) = 
 -4.2971e-09  1.4740e-01 -2.1405e-01  ...  -1.3160e-01 -3.5512e-04  1.3877e-01

( 12 ,.,.) = 
 -7.4684e-04 -4.1670e-01 -6.1688e-06  ...  -1.2905e-04 -2.9628e-05  3.0681e-02

( 13 ,.,.) = 
 -1.8500e-05 -5.9534e-04 -2.3576e-07  ...  -1.4994e-06 -2.1938e-05 -2.2080e-04
[torch.FloatTensor of size 14x1x1150]
, Variable containing:
( 0  ,.,.) = 
 -7.5986e-03  8.6798e-04 -6.7351e-03  ...   6.2188e-04  3.4969e-02 -9.1077e-02

( 1  ,.,.) = 
 -2.3921e-03  2.4249e-03 -1.7347e-03  ...   1.0938e-02 -3.6272e-04 -1.5099e-02

( 2  ,.,.) = 
 -2.7945e-02  9.5424e-04 -2.8550e-03  ...   9.9346e-05 -4.6923e-03 -1.5932e-02
 ... 

( 11 ,.,.) = 
  7.7081e-05  1.1043e-04  3.5880e-03  ...   3.4046e-03 -3.2344e-01 -7.6625e-04

( 12 ,.,.) = 
  3.6866e-03  3.5075e-03  7.5826e-04  ...  -2.6639e-02 -2.2028e-01 -2.5222e-03

( 13 ,.,.) = 
 -1.3276e-02  3.8389e-02  1.6594e-04  ...  -2.6982e-03 -1.2232e-01 -6.5385e-03
[torch.FloatTensor of size 14x1x1150]
, Variable containing:
( 0 ,.,.) = 
 -3.4815e-02  1.2264e-03  2.1081e-01  ...   1.3807e-01 -1.8228e-01 -4.8058e-03

( 1 ,.,.) = 
 -3.8202e-02 -1.0102e-01  1.4506e-01  ...   7.9288e-03 -1.2543e-01 -4.7196e-03

( 2 ,.,.) = 
 -2.1076e-02 -5.5695e-02  1.0110e-01  ...   9.5715e-03 -9.4924e-02 -1.6425e-03
... 

(11 ,.,.) = 
 -1.6449e-02 -8.2472e-04  3.1275e-02  ...   2.5666e-02 -9.2902e-02 -1.3512e-01

(12 ,.,.) = 
 -3.0238e-02 -2.3225e-03  1.2036e-01  ...   1.6667e-02  1.6427e-01 -2.0484e-02

(13 ,.,.) = 
 -2.2449e-01 -2.6965e-03  3.8396e-02  ...   1.3158e-03  4.0717e-02 -1.3653e-02
[torch.FloatTensor of size 14x1x400]
])
guten morgen schon novemberlicht da mag man gar nicht glauben was gestern wetter war - classification: [0]
Probability rating for sentence: Variable containing:
 0.4159  0.2991  0.2850
[torch.FloatTensor of size 1x3]