It’s weekend number two of the Deep Learning Specialization with the Purdue University Post Graduate Program in AI and Machine Learning program that I have been participating since January. This is the seventh and section of the program which won’t wrap up until mid-July with only the Capstone Project left to complete to finish out the program. This week-end was all about Object Detection, Selective Search, Sequential Modeling, Recurrent Neural Networks, Word Embedding, Long Short-Term Memory and Hybrid Modeling.
It’s all a bit like drinking from a firehose, as far as, all the information that is getting through at me, but learning and understanding more about the inter-workings of deep learning neural networks is why I wanted to enroll in this program. I say all that because my article list this time around is very much heavy on the AI/ML/DL topics.
AI/ML/DL/GenAI
A Grand Unified Theory of the AI Hype Cycle
I thought this was a good read and history lesson on the cyclical AI Hype Cycle.
Word Embedding
- Efficient Estimation of Word Representations in Vector Space
- Word2Vec Tutorial - The Skip-Gram Model
- Word Embedding Demo
- word_embeddings_demo.ipynb
- word-embeddings-workshop/Word Embeddings.ipynb
- The Illustrated Word2vec
Deep Learning
- Practical Deep Learning for Coders
- Deep Learning for Coders with fastai & PyTorch - The Book
- Understanding LSTM Networks
- LSTM Recurrent Neural Networks — How to Teach a Network to Remember the Past - Medium Membership Required to View Article
- A Visual Guide to Vision Transformers
GenAI
Spreadsheet Is All You Need - A nanoGPT pipeline packed in a spreadsheet
Haven’t had a chance to dig into this one yet, but sounds interesting. From the repo README: “This is a project that I did to help myself understand how GPT works. It is pretty fun to play with, especially when you are trying to figure out what exactly is going on inside a transformer. This helped me to visualize the entire structure and the data flow. All the mechanisms, calculations, matrices inside are fully interactive and configurable.” Dabo Chen