Clearly someone is bitter and leaving negative reviews on all Professor Peng's classes. She is great and I would highly recommend taking her class

The prof seems to care about her class, she responds very quickly on piazza, which I give her credits for. Unfortunately that's all the good things I can say about class. It actually pisses me more when the prof said in class the assignment won't be actually used in her research? Then why I am wasting my valuable time label those dumb datasets lol. Assignment 1 is not smart either. We are computer science majors, we do not like write a 5 page report just describe some random ChatGPT behaviors.

Please replace these assignments(also the final project) to some actual CODING project, such as write a transformer from scratch, GPT from scratch, translator from scratch.

Also it is pretty disappointing that the most advanced thing this class cover is transformers and some NLP jargons. EVERYBODY KNOWS TRANSFORMERS THESE DAYS. Please considering add more valuable content in your course.

Probably the best grad level course I have taken in UCLA! NLP area is not the same as ML/DL course but you'd better have some knowledge of DL before this course. This course cover traditional NLP techniques to the most advanced LLMs. Most assignments and final projects are research oriented. The content is also helpful if you are looking for jobs in AI/LLMs area. (They usually ask about transformer and attention in interview)

I think the name of this class should be "Introduction to NLP research". For someone who have background in the NLP/AI research, the content covered is pretty basic. Assignment 1, which is paper reading, is pretty natural to a person accustomed to research. Assignment 2, which I agree is a bit of a chore, probably could be streamlined with API calls to ChatGPT API. Maybe that's what research in NLP is like? I don't know. I feel like the comments about technical details are not doing justice to the professor. I think this class is detailed enough as a research oriented class. (Even for undergrad classes like 146 or 145, the professors are not covering code implementation in class. Implementation is practiced in the projects.) Since the assignments are designed to be research focused, student don't really have the chance to practice implementation of algorithms.
On the other hand, as a research oriented class, I feel that there is too few material that comes from current ongoing research. The current form of this class is somewhere in the middle of the two worlds(content oriented/research oriented), which misses the merit of both worlds.
Therefore, I think the professor need to readjust the focus of the class. Advertise the class as a research oriented class. Put more focus on current research. Leave the fundamental content/implementation project to 162. Make 162 a pre-requisite.

As for the course content, I believe it is quite fair. It covers a wide range of topics and explains the historical development of NLP. The content also balances concepts and mathematical explanations well. I personally think the course content is suitable for students with various backgrounds;

As for the quiz, I think it is unfair. The practice quiz should help us to better prepare for the actual quiz. However, the practice quiz is quite different from the actual one, which turns out to be misleading us. Also, some questions on the quizzes concentrate on unnoticeable details. Given what we have learned during the lecture, I feel like I am doing zero-shot or one-shot during the quizzes.

As for the assignments, the 1st one is fine. But the workload for the 2nd one is too overwhelming. And since the 1st assignment is already research-orientated, making the 2nd one even more research-orientated is unnecessary. I believe a better way is to have an application-orientated assignment for the 2nd assignment.

Good class overall. This class covers lots of useful information related to current NLP research. However, the workload is heavy. 2 midterms + 1 final + 2 big assignments + 1 final project seems too much. I have to admit that this workload is not as bad as it seems since assignments and the final project are actually not hard. But if we can cancel the final or one of the midterms, it will be more manageable.

Great overview of current NLP papers. Some NLP experience is necessary, but grading is very generous and mostly based on participation/completion. Not sure what the other reviews are saying since Prof Peng is really knowledgable, helpful, and completely fluent in English

Violet is such a sweet professor, and she truly cares about student learning. Her lectures really go in depth about NLP concepts, which was a bit overwhelming as someone who has no NLP experience, but she really tries to explain things in a way that is clear and easy to understand. She has two quizzes, which aren't too difficult (even though I didn't do great on them). The tricky part is that she really emphasizes certain aspect of the course, so you really need to make sure you have a good understanding of everything covered. The two assignments aren't too difficult, and I like that they introduced us to what NLP research looks like. Overall, she did a really great job, especially for her first time teaching the course.

Really great class for anyone with a little bit of NLP experience who wants to know more. Between homework, projects, and quizzes, there's good coverage of NLP basics, underlying math, and important existing research. Assignments are creative and go beyond just problem sets. Prof Peng and the TAs are super helpful and responsive. Lectures are engaging with a good amount of participation/dialogue with students. There are lots of extra credit opportunities.

The graduate version of the course:CS263
I still can't believe they split their research dataset to us to annotate it as our HW2. It was pathetic that they tried to fool us. Also HW1 was not helpful. They gave us a list of about 30 useless papers to read and make video for one of them.
She teaches in a very high level and shallow way and don't go through the details well.
Exams are always multiple choice tricky questions.
Project was also nothing but how to work with chatGPT!
I blame myself for taking this useless time-consuming course. This course has no good TA and lecturer at all.