NLP Foundations
How machines learned to read
From bag-of-words to BERT — the history and mechanics of natural language processing before (and leading to) the transformer era.
Intermediate3 modules · 9 lessons~72 min695 XPAligned with Hugging Face NLP CourseStart with: ML Foundations
Progress0/9
Module
Text as Data
Ambiguity, word counts, and the first language models
Why Text Is Hard7 min · 60 XPWhy Text Is Hard
Bag of Words & TF-IDF8 min · 70 XPBag of Words & TF-IDF
N-grams: The First Language Models8 min · 75 XPN-grams: The First Language Models
Module
Meaning as Vectors
Word2vec, GloVe, and the geometry of language
Words Become Geometry8 min · 80 XPWords Become Geometry
From Words to Whole Sentences7 min · 70 XPFrom Words to Whole Sentences
The NLP Task Zoo9 min · 85 XPThe NLP Task Zoo
Module
The Pre-Transformer Era
RNNs, seq2seq, and the bottleneck attention broke
Reading Word by Word: RNNs & LSTMs8 min · 80 XPReading Word by Word: RNNs & LSTMs
Seq2seq & the Bottleneck9 min · 85 XPSeq2seq & the Bottleneck
Why Attention Was Inevitable8 min · 90 XPWhy Attention Was Inevitable
Capstone project
Language under the hood
Build a text classifier — spam, sentiment, or topic — first with classic features (TF-IDF), then with embeddings. Compare the two approaches on the same test set.
Deliverable: A notebook showing both pipelines, their metrics, and what the embeddings captured that bags-of-words couldn't.
Milestones
Finish every lesson to earn the NLP Foundations Certificate.