Programming Assignment: Programming Assignment: Question Answering - C4W3 Many warning in unit test

Course Q&A Natural Language Processing NLP with Attention Models
1

I encountered many warnings in unit test. I unable to get the Expected output.

I hope you can help me on this. Thanks in advance.

2025-02-01_15-53-11
2025-02-01_15-53-111232Ă—1128 63.2 KB

2

I found this discussion on Stack Overflow:

Were you trying to load a model checkpoint to run prediction?

3

@chenkhoon

try to disconnect and reconnect the kernel, then run the cell from beginning till the unittest you encountered this output, if you still are getting same output, then let me know.

my also suspicion would be that you might have made changes in the assignment codes that you weren’t suppose to. in case you made any changes other than just writing codes with the already instructions given, then get a fresh copy and re-do your assignments. Please make sure only to write codes between the assigned markers ###START AND END CODE HERE### with instructions already given when you open the notebook.

4

I am implementing a “question-answering function using a Transformer model” in “TensorFlow”.

5

I have re-done the cell by missing the missing code as follows

it still show me the error.

Error
Error1224Ă—1243 76.4 KB

def answer_question(question, model, tokenizer, encoder_maxlen=150, decoder_maxlen=50):
A function for question answering using the transformer model
Arguments:
question (tf.Tensor): Input data with question and context
model (tf.keras.model): The transformer model
tokenizer (function): The SentencePiece tokenizer
encoder_maxlen (number): Max length of the encoded sequence
decoder_maxlen (number): Max length of the decoded sequence
Returns:
(str): The answer to the question

Codes removed, against community guidelines.

return tokenized_answer
6

By the way I also restarted the kernel and clear all output cell. Please advise me further on this.

7

don’t post codes here @chenkhoon.
kindly DM me screenshot of the codes by personal DM. click on my name and then message.

I noticed many of the codes are left with None. why have you not written codes completely??

Remember the codes given come with some instructions and already given part of the codes with None

those None need to be replaced with your codes, only time None need to be left when in the instructions it is mentioned to be left None as it is.

8

I did nott post the code. What I posted is the original code. Please advise how can I DM you my code?

9

I DM you my code already…

10

oh ok. :joy: smart

you can send me code by DM

11

@chenkhoon

Errors in your codes
for code line

  1. Predict the next word using the model, the input document and the current state of output

you are using a global variable for the model, remember for the answer_question code cell, the assigned variable name for model

model (tf.keras.model): The transformer model

  1. for code
    Concat the predicted next word to the output
    your axis is incorrect, remember you have to predict the next word and not the previous word.

  2. for code line
    The text generation stops if the model predicts the EOS token

your if statement should satisfy the below condition

Stop Condition:

The text generation stops if the model predicts the EOS token.
If the EOS token is predicted, break out of the loop.

if your next is absolute equal to eos, then break the loop.(you do not require to mention any other numpy or indexing for this code.

12

Thanks for the advice. Resolved with great thanks

1 Like