Lensa ML
Track 4 of 6

LLMs & Generative AI

Text representations, language models, transformers, diffusion, RLHF, LoRA, and RAG.

You want to understand how LLMs and generative models work inside. 4 paths · 22 explainers.

Path 01 · 7 explainers

NLP I: Text as Numbers

Tokenization, classic text features, and word embeddings

  1. 1Tokenization & BPE
  2. 2Bag of Words
  3. 3TF-IDF
  4. 4Word2Vec & Word Embeddings
  5. 5CBOW (Continuous Bag of Words)
  6. 6Skip-gram
  7. 7Embeddings
Path 02 · 5 explainers

NLP II: Language Models

Classification, seq2seq, NER, and language modeling with BERT

  1. 1Text Classification & Sentiment
  2. 2Seq2Seq & Beam Search
  3. 3Named Entity Recognition
  4. 4Language Modeling
  5. 5BERT & Masked Language Modeling
QKV
Path 03 · 6 explainers

How LLMs Work

From attention to decoder-only transformers, the architecture behind GPT

  1. 1How Self-Attention Works
  2. 2Multi-Head Attention
  3. 3Positional Encoding
  4. 4Cross-Attention
  5. 5The Transformer Architecture
  6. 6Decoder-Only Transformers
Path 04 · 4 explainers

Generative AI

Diffusion models, RLHF, LoRA, and retrieval-augmented generation

  1. 1Diffusion Models
  2. 2Retrieval-Augmented Generation
  3. 3LoRA & Parameter-Efficient Fine-Tuning
  4. 4RLHF