# Attention modelf **URL:** \ **Category:** Sequence Models\ **Tags:** coursera-platform\ **Created:** [September 1, 2021, 10:07am UTC](https://community.deeplearning.ai/t/attention-modelf/32432 "2021-09-01T10:07:30Z")\ **Posts on this page:** 7\ **Page:** 1
**Author:** ![Igba](https://avatars.discourse-cdn.com/v4/letter/i/0ea827/32.png) [@Igba](https://community.deeplearning.ai/u/Igba)\ **Post date:** [September 1, 2021, 10:07am UTC](https://community.deeplearning.ai/t/attention-modelf/32432/1 "2021-09-01T10:07:30Z")
Kindly help my code in week 3 assignment for attention Modelf is returning the following error; ValueError Traceback (most recent call last) in 34 35 —\> 36 modelf\_test(modelf) in modelf\_test(target) 11 12 —\> 13 model = target(Tx, Ty, n\_a, n\_s, len\_human\_vocab, len\_machine\_vocab) 14 15 print(summary(model)) in modelf(Tx, Ty, n\_a, n\_s, human\_vocab\_size, machine\_vocab\_size) 37 38 # Step 2.A: Perform one step of the attention mechanism to get back the context vector at step t (≈ 1 line) —\> 39 context = one\_step\_attention(a[t],s[t-1]) 40 41 # Step 2.B: Apply the post-attention LSTM cell to the “context” vector. in one\_step\_attention(a, s\_prev) 17 ### START CODE HERE ### 18 # Use repeator to repeat s\_prev to be of shape (m, Tx, n\_s) so that you can concatenate it with all hidden states “a” (≈ 1 line) —\> 19 s\_prev = repeator(s\_prev) 20 # Use concatenator to concatenate a and s\_prev on the last axis (≈ 1 line) 21 # For grading purposes, please list ‘a’ first and ‘s\_prev’ second, in this order. /opt/conda/lib/python3.7/site-packages/tensorflow/python/keras/engine/base\_layer.py in **call** (self, \*args, \*\*kwargs) 924 if \_in\_functional\_construction\_mode(self, inputs, args, kwargs, input\_list): 925 return self.\_functional\_construction\_call(inputs, args, kwargs, → 926 input\_list) 927 928 # Maintains info about the `Layer.call` stack. /opt/conda/lib/python3.7/site-packages/tensorflow/python/keras/engine/base\_layer.py in \_functional\_construction\_call(self, inputs, args, kwargs, input\_list) 1090 # TODO(reedwm): We should assert input compatibility after the inputs 1091 # are casted, not before. → 1092 input\_spec.assert\_input\_compatibility(self.input\_spec, inputs, self.name) 1093 graph = backend.get\_graph() 1094 # Use `self._name_scope()` to avoid auto-incrementing the name. /opt/conda/lib/python3.7/site-packages/tensorflow/python/keras/engine/input\_spec.py in assert\_input\_compatibility(input\_spec, inputs, layer\_name) 178 ‘expected ndim=’ + str(spec.ndim) + ‘, found ndim=’ + 179 str(ndim) + '. Full shape received: ’ + → 180 str(x.shape.as\_list())) 181 if spec.max\_ndim is not None: 182 ndim = x.shape.ndims ValueError: Input 0 of layer repeat\_vector is incompatible with the layer: expected ndim=2, found ndim=1. Full shape received: [64] ---
**Author:** ![Kic](https://sea1.discourse-cdn.com/dlai/user_avatar/community.deeplearning.ai/kic/32/76339_2.png) [@Kic](https://community.deeplearning.ai/u/Kic)\ **Post date:** [September 1, 2021, 11:55am UTC](https://community.deeplearning.ai/t/attention-modelf/32432/2 "2021-09-01T11:55:05Z")
Hi @Igba The parameters passed to one\_step\_attention() are incorrect. It should be just a, output of the bi-directional LSTM, and s, previous state. a is sequence of hidden states, and s is initialised to s0, and t is the timestep. At each time step, t, until Ty, the same sequence of a is used, but s is updated after calling post\_activation\_LSTM\_cell(). ---
**Author:** ![Igba](https://avatars.discourse-cdn.com/v4/letter/i/0ea827/32.png) [@Igba](https://community.deeplearning.ai/u/Igba)\ **Post date:** [September 2, 2021, 8:11pm UTC](https://community.deeplearning.ai/t/attention-modelf/32432/3 "2021-09-02T20:11:21Z")
Thanks sir, but i am still getting same error. I sent you a private message of my code to vet please. ---
**Author:** ![Kic](https://sea1.discourse-cdn.com/dlai/user_avatar/community.deeplearning.ai/kic/32/76339_2.png) [@Kic](https://community.deeplearning.ai/u/Kic)\ **Post date:** [September 2, 2021, 8:47pm UTC](https://community.deeplearning.ai/t/attention-modelf/32432/4 "2021-09-02T20:47:07Z")
Hi @Igba Two problems with your call to Bidirectional LSTM: 1. wrong input units, it should be : units=n\_a , not 1, because n\_a – hidden state size of the Bi-LSTM (see comments) 2. input shape is missing, it should be : input\_shape=(m,Tx,n\_a\*2) ---
**Author:** ![Igba](https://avatars.discourse-cdn.com/v4/letter/i/0ea827/32.png) [@Igba](https://community.deeplearning.ai/u/Igba)\ **Post date:** [September 3, 2021, 5:02am UTC](https://community.deeplearning.ai/t/attention-modelf/32432/5 "2021-09-03T05:02:04Z")
> [@Kic](#): > > input\_shape=(m,Tx,n\_a\*2) Thanks sir, I did this : a = Bidirectional(LSTM(units= n\_a, return\_sequences= True),input\_shape=(m,Tx,n\_a\*2)) but i still got this : AttributeError: ‘Bidirectional’ object has no attribute ‘shape’ ---
**Author:** ![Kic](https://sea1.discourse-cdn.com/dlai/user_avatar/community.deeplearning.ai/kic/32/76339_2.png) [@Kic](https://community.deeplearning.ai/u/Kic)\ **Post date:** [September 3, 2021, 10:32am UTC](https://community.deeplearning.ai/t/attention-modelf/32432/6 "2021-09-03T10:32:44Z")
you missed the (X) at the end. ---
**Author:** ![Igba](https://avatars.discourse-cdn.com/v4/letter/i/0ea827/32.png) [@Igba](https://community.deeplearning.ai/u/Igba)\ **Post date:** [September 4, 2021, 4:35am UTC](https://community.deeplearning.ai/t/attention-modelf/32432/7 "2021-09-04T04:35:02Z")
Many thanks sir, it worked.