C5_W4_A1_Transformer_Subclass_v1 UNQ_C6 Wrong values

Course Q&A Deep Learning Specialization Sequence Models
1

Hi all,

I’ve already looked on this forum for an answer. But I’m stuck at

UNQ_C6 (UNIQUE CELL IDENTIFIER, DO NOT EDIT)

GRADED FUNCTION DecoderLayer

class DecoderLayer(tf.keras.layers.Layer):

attn_w_b1 and attn_w_b2 are both correct (no error on assert)
but it end with:
→ 182 assert np.allclose(out[0, 0], [-0.22109576, -1.5455486, 0.852692, 0.9139523]), “Wrong values in out”

AssertionError: Wrong values in out

My values of out[0,0]
print(out3[0,0])
tf.Tensor([-1.7139359 0.7697465 0.36217463 0.58201486], shape=(4,), dtype=float32)

I’ve checked the described steps of block 3 and don’t see any error in my code. but there must be on. Can I send my code (private message of course) to someone, please?

2

Hey @Erwin_van_Dijk,
Welcome back to the community. Please DM me your code for this function, so that we can figure out where exactly does the error reside in your code.

Cheers,
Elemento

3

Sometimes it is good, to start from scratch again and compare with the precious attemt.
I should have used Q1 for the sum, but I had used mult_attn_out1. I didn’t see it until I started with a fresh and empty assigment.

1 Like