Full Stack Deep Learning - 2022 Course
All the lecture and lab material is free forever. Just check out the links below.
We released lecture videos on Mondays at 6pm Pacific and lab videos on Wednesdays at 6pm Pacific on YouTube.
|2022.07.25||Pre-Labs 1-3: CNNs, Transformers, and PyTorch+Lightning||-|
|2022.08.08||Lecture 1: Course Vision and When to Use ML||Lab Overview||-|
|2022.08.15||Lecture 2: Development Infrastructure & Tooling||Lab 4: Experiment Management||-|
|2022.08.22||Lecture 3: Troubleshooting & Testing||Lab 5: Troubleshooting & Testing||-|
|2022.08.29||Lecture 4: Data Management||Lab 6: Data Annotation||Start forming groups|
|2022.09.05||Lecture 5: Deployment||Lab 7: Web Deployment||Group proposals due|
|2022.09.12||Lecture 6: Continual Learning||Lab 8: Model Monitoring||Work on project|
|2022.09.19||Lecture 7: Foundation Models||Work on project|
|2022.09.26||Lecture 8: ML Teams and Project Management||Work on project|
|2022.10.03||Lecture 9: Ethics||Work on project|
|2022.10.10||Project Presentations||Project due|
We review some prerequisites -- the DNN architectures we'll be using and basic model training with PyTorch -- and introduce PyTorch Lightning. Published August 10, 2022.
We review the purpose of the course and consider when it's a good (or bad!) idea to use ML. Published August 8, 2022.
We walk through the entire architecture of the application we will be building, from soup to nuts. Published July 25, 2022.
We tour the landscape of infrastructure and tooling for developing deep learning models. Published August 15, 2022.
We run, track, and manage model development experiments with Weights & Biases. Published August 17, 2022.
We look at tools and practices for testing software in general and ML models in particular. Published August 22, 2022.
We try out some Python testing tools and dissect a PyTorch trace to learn performance troubleshooting techniques. Published August 24, 2022.
We look at sourcing, storing, exploring, processing, labeling, and versioning data for deep learning. Published August 29, 2022.
We spin up a data annotation server and learn just how messy data really is. Published August 31, 2022.
We do a lightning tour of all the ways models are deployed and do a deep dive on running models as web services. Published September 5, 2022.
We create and deploy our ML-powered text recognition application with a simple web UI and a serverless model service. Published September 7, 2022.
We consider what it takes to build a continual learning system around an ML-powered application. Published September 12, 2022.
We add user feedback to our ML application and review data logged by actual users of the FSDL Text Recognizer. Published September 14, 2022.
We look at how to build on GPT-3, CLIP, StableDiffusion, and other large models. Published September 19, 2022.
We look at the structure of ML teams and projects, including how to hire or get hired on an ML team and how to build an ML-first organization. Published September 26, 2022.
We consider ethical concerns around buiding technlogy, building with machine learning, and building artificial intelligence. Published October 3, 2022.
This course was only possible with the support of our amazing TAs (in alphabetical order):
- Andrew Mendez is a Deep Learning Solutions Engineer at DeterminedAI, working on computer vision and NLP solutions for defense and autonomous vehicle companies. Previously Andrew worked as an ML Engineer at Clarifai and CACI.
- Daniel Hen is a Senior Data Scientist at Digital Turbine, working on Ad Tech and mobile solutions, as well as Big Data problems. Working with Spark, ML algorithms such as XGBoost, Computer Vision, and constantly learning new technology.
- James Le runs Data Relations and Partnerships at Superb AI, a data management platform for computer vision use cases. Outside work, he writes data-centric blog posts, hosts a data-focused podcast, and organizes in-person events for the data community.
- Saurabh Bipin Chandra is a Senior ML Scientist at Turnitin.
- Sayak Paul is as a Machine Learning Engineer at Carted on NLP and representation learning from HTML webpages. Besides work, he contributes to various open-source projects.
- Vishnu Rachakonda is a Data Scientist at firsthand.
We are excited to share this course with you for free.
We have more upcoming great content. Subscribe to stay up to date as we release it.
We take your privacy and attention very seriously and will never spam you. I am already a subscriber