Is there anything for the student to actually code or fill in for the optional labs for Course 5, Week 4 (Transformers)? It’s worded confusingly; in the second optional lab there is a section labeled " Exercise 1 - tokenize_and_align_labels", but the function seems to be already fully filled in. And the rest of the labs have lots of wording with “you will…” do such and such, but again, the relevant sections seem all filled in, and I can’t find any “Your code here”-type comments within the code. Can anyone clarify?
You probably referring to ungraded labs which aims for learners to understand about how a specific topic assignment codes works.
for graded assignment it usually present after the quiz page or at the end of week for you to write codes and pass the assignment.
These labs are more guided and demonstration-based. In these labs, the functions are usually already implemented, and the intent is more for learners to study how the code works, understand the logic behind the functions and experiment if they’d like to tweak or extend the code provided.
If you’re looking to get more hands-on experience:
- You can try modifying the code to experiment with different approaches or parameters.
- You might try to extend the functions, such as adapting them for different use cases, changing tokenizers, or working with different models.
This approach encourages a more advanced level of engagement, where you’re expected to understand and critique the code provided, rather than focusing on implementation from scratch.
1 Like
Thank you for your response; yes, I am referring to the ungraded labs. Specifically: “Lab: Transformer Preprocessing”, “Lab: Transformer Network Application: Named-Entity Recognition”, and “Lab: Transformer Network Application: Question Answering”. I understand that they are ungraded, and therefore would not have any formal graded “exercises”; however, I believe that other optional labs in the Deep Learning specialization do still have gaps in the code for the student to fill in in order for the cells to give the expected results. I was wondering if these three optional labs had such gaps, or if as it seems to me, all the code is completely filled in for these labs.
ungraded labs don’t have these gaps but as far as I hope you are talking about DLS course only. NER, question answer assignment seemed NLP assignment names. I hope you have posted your query in the right category of specialisation.
Ungraded labs is only for your understanding on how taught lesson for the week are programmed as models, so you use or understand these labs for you to do to better in graded assignment.
I am definitely referring to labs in the DLS; Course 5: Sequence Models, Week 4: Transformers. The lab names I gave are the exact names of the optional labs for Week 4 of that course, in the Deep Learning Specialization.
1 Like
okay it’s been months I haven’t opened DLS course as I am NLP mentor the assignment name seemed similar, so was just making sure query is in the right category. Yes I remember transformer week 4 assignment in DLS it explains the basics of attention mechanism. it’s a fun exercise, do practice!
1 Like
You do not have to add code to the optional labs. They are more like demonstrations.
Thank you for the clarifications, everyone who responded. I still think for future reference, the instructions and introductions to the sections in these labs should be made clearer, to make it clear that there is nothing that the student has to implement in order for the functions to work at a basic level, so students aren’t searching for the sections where they will have to make edits. Phrases like “You will implement”, which occurs in one of the labs, strongly imply to me that such edits are required.