If you get an error in the EncoderLayer test case about “wrong values”, check your code for this issue:
When applying layernorm1(), the instructions mention “attention output”, and there it refers to the “attn_output” variable.
But when applying layernorm2(), the instructions refer to “output from multi-head attention”, but there it means “out1”.
what does input refer at that step? is not it x?
I am getting error “wrong values”
AssertionError Traceback (most recent call last)
in
17
18
—> 19 EncoderLayer_test(EncoderLayer)
in EncoderLayer_test(target)
12 [[-0.5214877 , -1.001476 , -0.12321664, 1.6461804 ],
13 [-1.3114998 , 1.2167752 , -0.5830886 , 0.6778133 ],
—> 14 [ 0.25485858, 0.3776546 , -1.6564771 , 1.023964 ]],), “Wrong values”
15
16 print("\033[92mAll tests passed")
AssertionError: Wrong values
The instructions in section 4.1 tell you that the input is ‘x’.
I was doing wrong on other step.
Thank you