## Learning Objectives
The goal of the course is to introduce deep learning from a computer science perspective. Instead of proving properties, we will be focusing on a practical understanding of neural networks, training, and inference techniques. Students will learn to design and train networks using modern deep learning toolkits such as PyTorch, Huggingface. Course topics include, but are not limited to MLPs, CNNs, RNNs, Transformers (including LLMs), Diffusion Models, and distributed training.
If there is any topic you are particularly interested in, feel free to reach out to me.
### Prerequisites
You should be very familiar with Python. You should have taken a significant subset of Calc I, II, III, Prob + Stats, and Linear Algebra.
## Computing Platform
We will be training and evaluating many models throughout the course and using Jupyter Notebooks. You can use Colab or other platforms. You may find a paid subscription to access GPUs / TPUs may be useful. You also have access to an OpenHPC, an education cluster, where you can run your models.
## Grading
The tentative grading breakdown will be:
Final Project Proposal
15%
Final Project
20%
Assignments
60%
Class Participation + Attendance
5%
### Final Project
{% hint style="danger" %}
This is a tentative plan and may be replaced by a final exam
{% endhint %}
The current plan is to do a significant end-of-semester project where students are expected to plan and produce a significant end product. In general, the topic and goal of the project are open-ended, but must include substantial data wrangling, training, or fine-tuning of models, and proof of significant achievement.
I will work with you on the project proposal and ensure you can complete your plan within the semester. The plan is for you to work with me through the semester to refine your idea, implement it, and gather results.
You should expect to give a 5-10 minute talk on your project and write a 4-8 page (including references) report. I hope these projects will spur your creativity and showcase what you've learned over the semester.
### Programming Assignments
We will use GitHub Classrooms and Piazza. Assignments will be due at 11:59 PM the day they are due. Assignments will lose 7% for each day they are late, unless explicitly granted an extension. Extensions are difficult for the shortened Summer semester.
### Class Participation
The course is synchronous and attendance is mandatory, and roll will be taken. If you are late to joining the class, ping me on Piazza. Due to the shortened semester, ping me to ask for excused absences.
In general, I also encourage participation and discussion on the class forum on Piazza.
### Academic Integrity
You are free to discuss assignments with one another at a high level. However,
* You must write your code.
* You must not show your code to other students.
* You must not obtain code from AI.
* Do not look at other students' code.
* Be sure to always log out when you leave the lab.
* Do not share your laptop with other students.
* Limited borrowing of code from the Internet and other sources is usually allowed, but you must okay it first with the instructor; and you must cite it.
Students submitting programming assignments or exams that we decide are too similar, will receive no credit on that assignment, and the maximum grade those students can receive for the class is a C. We will be using source code comparison programs to check programs against one another. Furthermore, your GitHub commits must reflect that you actually worked on the program. Further details as to the requirements will be forthcoming.
---
# Agent Instructions: Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.
Perform an HTTP GET request on the current page URL with the `ask` query parameter:
```
GET https://szaman.gitbook.io/intro-to-deep-learning/summer-2025-course-details.md?ask=
```
The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.
Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.