Open mli42 opened 2 years ago
Given examples of loss_elem_ are 10 times greater than it should. \ The subject says:
loss_elem_
# Example 0.1: lr1.loss_elem_(y, y_hat) # Output: array([[710.45867381], [364.68645485], [469.96221651], [108.97553412], [299.37111101]])
But the loss is "only" at 195...
# Example 0.2: lr1.loss_(y, y_hat) # Output: 195.34539903032385
With my code, I got this for loss_elem_ and I have the right loss_:
loss_
array([[71.04586738], [36.46864549], [46.99622165], [10.89755341], [29.9371111 ]])
The same issue is repeted for lr2, subject loss_elem_:
lr2
# Example 1.2: lr2.loss_elem_(y, y_hat) # Output: array([[486.66604863], [115.88278416], [ 84.16711596], [ 85.96919719], [ 35.71448348]])
Subject loss:
# Example 1.3: lr2.loss_(y, y_hat) # Output: 80.83996294128525
My code:
array([[48.66660486], [11.58827842], [ 8.4167116 ], [ 8.59691972], [ 3.57144835]])
Given examples of
loss_elem_
are 10 times greater than it should. \ The subject says:But the loss is "only" at 195...
With my code, I got this for
loss_elem_
and I have the rightloss_
:The same issue is repeted for
lr2
, subjectloss_elem_
:Subject loss:
My code:
Fixed on: