CMU10-714

---

Author: Shihong Yuan Sopheremore

Beginning Time: 2024/10/12

---

1 Inspiration:

1. I've been working on front-end and back-end since summer, at first, I thought these two ends will make it clearer that how all the computer end to end projects are made. But after getting into the door, building teo basic projects, I felt little motivation to keep on learning as I dthought I need more same level friends to talk with or say being taught, also I need a great project chance to finish( Because I think it's can only by chance gain little improvement). That's why I love American class as they offer gradually step lab and slides. That little improvemnt are not by chance, but by efforts.
2. UIUC has a new but powerful professor that I really would like to come up with him, and his career is about MLsys, which inspire me the most to study on this.
3. Also I am working on ECE391@uiuc to build a linux like operating system, which seemingly a lot differnce with this course.\
4. Inspired by zhihu post https://zhuanlan.zhihu.com/p/632617197

2 What is here:

1. slide(code also if have) & notes & book
2. hw
3. final project

3 What will I do:

1. I will finish this in 15 days. 11.20 night to 11.30 night for example.Except the project (Becuase I know most of the theory knowledge before.)
2. Then I can work on my linux ece391@uiuc in about 10 days.(Becuase I have been working on that about 10 days yet and environments can just use wsl which I tried a lot time weeks ago) To about 11.10
3. Then Parallel Computing CMU 15-418/Stanford CS149. For about 10 days(Because I didn't know it well)
4. Communicate with seniors Ni and know much about ros, only using one weekend all in the lab.(Because long time makes effect.)
5. After that, I could connect to the professor and to see whether can I follow him.(The ddl must before 11.15)

4 Reference[I don't really know whether will this break the honr code, but whatever learning fully and effectively is the most important.]

1. https://github.com/PKUFlyingPig/CMU10-714
2. https://www.zhouxin.space/notes/notes-on-cmu-10-414-assignments/(I notice this from zhihu)
3. https://www.zhouxin.space/notes/notes-on-cmu-10-414-deep-learning-system/

5 A small conclusion for the course[waiting](Content by gpt )

Video Resources: The course videos are available on Bilibili and are suitable for beginners to learn deep learning systems. The lectures start with simple classification problems and backpropagation optimization and gradually cover more complex neural networks.

Homework and Labs: The course includes six assignments (homework) that guide students through implementing a deep learning framework from scratch, including CNNs, RNNs, Transformers, and other common neural network models.

Course Introduction:

Prerequisites: Basic knowledge of systems programming, linear algebra, probability, Python, C++, and some experience with machine learning. Suitable Learning Stage: Suitable for students who have a certain foundation in machine learning and want to delve into the underlying implementations of deep learning frameworks. Course Evaluation: The course is challenging, with high-quality lectures and homework that is both difficult and of high quality, providing a deep understanding of deep learning systems. Non-official Quality Course Resources: There are study notes and translated course content available, which can help understand the course material. Follow-up Course Recommendations: The course is comprehensive and stands alone, but for further study, you might consider more advanced topics in deep learning or specialized applications. Non-official Quality Course Resource Links: Such as notes, course content translations, etc., can be found on platforms like CSDN or personal blogs where students share their study notes and experiences.

Subsequent Course Recommendations: For those interested in further study, there are many advanced courses and specializations in deep learning that build on the knowledge gained from CMU 10-414.