afshinea/stanford-cs-230-deep-learning

★ 6,981Audience · researcherComplexity · 1/5Setup · easy

Mindmap

mindmap
  root((repo))
    What it does
      Deep learning reference
      Stanford CS 230 summary
    Topics covered
      Convolutional networks
      Recurrent networks
      Training tips and tricks
    Format
      PDF cheatsheets
      Multi-language versions
      Website version available
    Audience
      Students
      Researchers
      Practitioners

mindmap root((repo)) What it does Deep learning reference Stanford CS 230 summary Topics covered Convolutional networks Recurrent networks Training tips and tricks Format PDF cheatsheets Multi-language versions Website version available Audience Students Researchers Practitioners

Click or tap to explore — scroll the page freely

Things people build with this

USE CASE 1

Review key deep learning formulas and architecture diagrams before an exam or technical interview.

USE CASE 2

Share with teammates as a shared quick-reference during a deep learning project or reading group.

USE CASE 3

Look up how CNN output dimensions are calculated or how LSTM gates work without rereading a textbook.

Getting it running

Difficulty · easy Time to first run · 5min

In plain English

This repository contains cheatsheets for Stanford's CS 230, a university course on deep learning. Deep learning is the branch of artificial intelligence that trains multi-layered neural networks on large amounts of data to recognize patterns, generate text or images, translate languages, and many other tasks. CS 230 is one of Stanford's most well-known AI courses. The cheatsheets distill the main concepts from the course into compact reference documents in PDF format. There are three focused cheatsheets: one on convolutional neural networks (a type of model commonly used for image recognition and computer vision tasks), one on recurrent neural networks (models designed to handle sequences, such as text or audio), and one on general tips and tricks for training deep learning models. A fourth document combines all three into a single summary reference. All of the cheatsheets are available in multiple languages, including English, Persian, French, Japanese, Korean, Turkish, and Vietnamese. The same material is also hosted on a website, which makes it easier to read on phones and tablets without downloading the PDF files. The cheatsheets were created by two researchers, Afshine Amidi and Shervine Amidi, affiliated with Ecole Centrale Paris, MIT, and Stanford University. A companion repository is available if anyone wants to contribute translations into additional languages. This is a reference and study resource, not a software project. There is no code to run. It is most useful to someone who is taking or has taken a deep learning course and wants a concise summary of the core concepts to review or look up quickly.

Copy-paste prompts

Prompt 1

Using the Stanford CS 230 CNN cheatsheet as a reference, explain how to calculate the output width and height after a convolution layer with a 3x3 filter, stride 2, and same padding.

Prompt 2

Based on the CS 230 deep learning tips cheatsheet, walk me through the recommended diagnostic steps to identify and fix overfitting in a model I am training.

Prompt 3

I am implementing an LSTM for text generation. Using the CS 230 RNN cheatsheet, explain the forward pass equations and show how the forget, input, and output gates prevent the vanishing gradient problem.

Open on GitHub → Explain another repo

← afshinea on gitmyhub — every repo by this author, as a profile.

Verify against the repo before relying on details.