C4_W1_UNQ_C6 wrong ouput

Hi all, I’m having the following errors while testing the function next_symbol:

I’ve tried to follow the other closed issues for this part of the assignment, anyway I’m not able to get a correct probability output.
Moreover, if I try the sampling_decode() function (UNQ_C7) the model will never output the EOS token but it will stuck on a different token id.

Any help will be appreciated.
My id is : rvxrgtzxqfjp

Hey @Giovanni_Gazzola,
Welcome, and we are glad that you could be a part of this community! It’s hard to pin-point the exact issue without taking a look at your code in this case, so can you please DM me your implementation for the function. For DM, click on my name and select “Message”.

P.S. - For future references, don’t add your Lab ID, unless someone asks you for that explicitly.

Cheers,
Elemento

Hey @Giovanni_Gazzola,
Your formulation for padded is incorrect. Note that padded_length denotes the length of output tokens after appending zeroes. However, you have assumed that padded_length only denotes the number of zeroes. This leads to you appending an incorrect number of zeroes, and hence, the error. I hope this helps.

Cheers,
Elemento

Hi @Elemento,

you’re right! What a stupid error!
Thanks for the help and the fast reply.

Cheers,
Giovanni