QMSS GR5052

We worked on projects in NLP and Image Processing, in collaboration with the IBM Data Science Elite team. We really covered a lot of ground here, but I appreciated that it was extremely self-directed, and we were allowed to go as fast as we wanted to; we ended up working on very advanced material, including topic models that weren't covered on most class syllabi, and state of the art contrastive learning models.

how to train dataset in various LLM and benchmark to evaluate.

Nearly nothing compared to other courses

I didn't learn a lot from this course. I learned a lot from my peers and self-learning.

The course teaches nothing and the professor seems just filling her working hours.

I learned about how to research on my own and how to be proactive when working in a group that can be difficult to work with based off of language barriers.

This course was a joke... We have not receive the database (that we were supposed to work on during the semester) only in the last three weeks... We did not get any assistance or training. If you are not part of the core team the whole course is useless. That only thing that I have learned is how to google everything. 1 Columbia University: Arts & Sciences Fall 2021 Course: QMSSGR5052_001_2021_3-PRACTICUMINDATAANALYSIS : QMSSGR5052_001_2021_3 - PRACTICUM IN DATA ANALYSIS Instructor: Aracelis Torres

This course is supposed to offer a real-world project experience. I think the effectiveness of this structure is largely dependent on the team and client you are assigned to.

New coding packages, a lot of new industry-specific knowledge and background, how to analyze and interpret results on data that perhaps I do not understand but those who read the findings will. More familiarity with python. How to manage disappointing team dynamics in a grad-level course

100% student run/managed, groupwork based course. How much you enjoy this course depends on the projects available and mostly what your group is like - and its generally nothing amazing