Words to vectors, attention and the transformer, DistilBERT, Hugging Face pipelines, embeddings, and a semantic search engine.
Before attention can let words look at each other, each word has to become numbers. A math-free tour of vocabularies, token IDs, and the embedding lookup — built on a 12-word toy vocabulary you can run in PyTorch.
A gentle, math-free tour of the one idea behind every transformer — attention — and how it solves what convolution can't: letting far-apart words inform each other. Build and print a transformer block in a few lines of PyTorch.
Open a famous language model — DistilBERT — and read it like a spec sheet: the embeddings table, six stacked attention blocks, sixty-six million parameters, and named weights, all with the same nn.Module skills you used on ResNet. Math-free and beginner-friendly.
A beginner-friendly bridge from image preprocessing to text tokenization: token IDs, attention masks, padding, truncation, batching, and how to choose the right model input abstraction.
A beginner-friendly tour of the Hugging Face transformers pipeline — the fastest way to run a pretrained model in three lines, across text, images, and audio.
The text-classification capstone: turn pipeline('sentiment-analysis') into a real tool. Batch a pile of feedback through DistilBERT, route low-confidence cases to a human, and wrap it in a Gradio app with a sortable table and a live distribution.
A beginner-friendly guide to dropping below pipeline() for control — using AutoTokenizer and AutoModel from_pretrained, a huggingface tokenizer, and reading raw logits by hand.
A practical introduction to embeddings for PyTorch and Hugging Face users: turning text into vectors, comparing meaning with cosine similarity, batching inputs, and knowing when embeddings beat generation.
The transformer capstone: turn the embeddings from the last post into a tiny search engine that finds results by meaning, not keywords — embed a corpus, rank by cosine similarity, and wrap it in a Gradio app.