This is the Curriculum for Learn Machine Learning in 3 months (PyTorch Curriculum) by Siraj Raval on Youtube. Beginners to Python will learn to build, train, deploy, scale & maintain modern Machine learning & Deep learning models. Each weekly assignment will teach you how to use a new concept or tool, like Docker, PyTorch, or Transformer Models. The Final Project will integrate everything you've learned into a Self Driving Car simulation. After completion, start an ML startup or find relevant work in the field. Together as a learning community, we're going to help each other succeed!
- 🤝 Social: Join our Discord channel to find a study buddy
- ✨ Interactive: Every resource is web-based with user input
- 🧑🎓 Beginner-Friendly: Build weekly projects without dependencies thanks to codespaces
- 🤖 Project-Based: Learn Computer Vision, Natural Language Processing, Time Series Forecasting, Audio Processing, & Recommender Systems
- Python, Pip, Numpy, Pandas, Seaborn, Matplotlib, PyTorch, Replit, SQL, Jupyter, Streamlit, Gradio, HuggingFace, Airflow, GCP, AWS, Spark, Scikit-learn, Prometheus, Evidently, Grafana, Flask, Prefect, MongoDB, Postgres, Kafka, Terraform, RL-Baselines, Unity, W&B, Kubernetes, DBT
- Elicit to answer questions
- ExplainPaper to explain math
- Summari to explain text
- Spaces to sample demos
- CoPilot to explain code
Week 1: Python Fundamentals (Allen Downey)
Assignment: Build a Python search function for Researchers. Given a list of search terms, return a list of pages sorted by relevancy. Modify the example with your own alpha parameter.
Week 2: Mathematics of Machine Learning (xaktly.com)
Assignment: Solve the Bayesian probability problem for Supply Chain using pencil & paper. Do so after completing each full section on Calculus, Probability, Statistics, & Matrices.
Week 3: Data Analysis (Kaggle)
Assignment: Build a data visualization iPython notebook for Farmers. Search Kaggle for an agricultural dataset, then visualize it 3 different ways for comparison & further analysis.
Week 4: Machine Learning Techniques (Cyrille Rossant)
Assignment: Build a Random Forest Regression model for Real Estate. Clean, augment, & feature engineer a dataset to predict the price of houses next year in Boston
Week 1: Neural Networks (Interactive Dive into Deep Learning Book)
Assignment: Build a simple feedforward neural network for Retail. Upload the Jupyter Notebook to Colab, modify input data, monitor how it effects accuracy
Week 2: Transformers (HuggingFace Course)
Assignment: Build a conversational transformer for Mental Health therapy. Specifically, train Mini-GPT to have a therapeutic conversation by uploading it to Colab for training.
Week 3: Diffusers (Fast.AI Course)
Assignment: Build a design generator for Architects. Create a HuggingFace Space, select an existing image dataset, & create a web interface to generate designs.
Week 4: Deep Reinforcement Learning (Simonini Thomas)
Assignment: Train a Humanoid Robot to walk in simulation within a Jupyter Notebook for Construction projects. Generate a 10 second video of the humanoid walking.
Week 1: Design (Made with ML Course)
Assignment: Design a Medical Imaging Classification app for Doctors. Create the product requirements, design documentation, & project plan.
Week 2: Development (Full Stack Deep Learning Course)
Assignment - Package a pretrained text recognition model into a TorchSript binary, wrap it in a serverless cloud function, & build a simple UI.
Week 3: Production (DataTalks.CLub ML Ops ZoomCamp)
Assignment - Deploy a pretrained model for Traffic Prediction. Generate a report that detects any feature drift between model versions.
Week 4: Data Engineering (DataTalks.CLub Data Engineering ZoomCamp)
Final - Deploy a Self Driving Car Simulation app. This Javascript example is a great starting point. Integrate NLP, Computer Vision, Reinforcement Learning, & ML Ops.