ERROR in week 3 Machine Translation

[Yumlembam]

Hi,

In the line
----> 8 prediction = model.predict([source, s0, c0])

why am i getting the error "Error when checking : expected X to have 3 dimensions, but got array with shape (37, 30)"

[piyushsoni27]

Is there any resolution for this error yet? I am facing similar error. I am using Keras- "2.2.4"

[esraahisham753]

I have the same error also

[shanghaiJ]

I edited code with this:

source = string_to_int(example, Tx, human_vocab)
source = np.array(list(map(lambda x: to_categorical(x, num_classes=len(human_vocab)), source)))
ttt=np.expand_dims(source,axis=0)
print(ttt.shape)
prediction = model.predict([ttt, s0, c0])

Then, I got another error:
ValueError: Data cardinality is ambiguous:
x sizes: 1, 10000, 10000
Please provide data which shares the same first dimension.

I downgraded TF version to 2.0.0 instead of 2.3.0.

Anybody able to solve the issue? I am stuck on this.

[alfredomaussa]

i changed this line of code and worked:

prediction = model.predict([np.einsum('ij->ji', source).reshape((1,30,37)), s0[:1], c0[:1]])

[ai-phd-dream]

this also work
source = np.array(list(map(lambda x: to_categorical(x, num_classes=len(human_vocab)), source)))
prediction = model.predict([source.reshape(1,30,37), s0[:1], c0[:1]])