NVIDIA · NCA-GENL

The control thread

Steering what the model does and containing what it might do wrong: prompting as the first lever of control, RAG as grounding control, and trustworthy AI as the guardrails and audits that catch what neither lever stops.

T5 · 32 lessons across 4 modules

  1. M0505-01Zero-shot vs few-shot prompting and in-context learning
  2. M0505-02How to structure a prompt: instruction, context, and format
  3. M0505-03Chain-of-thought prompting: when it helps and when it misleads
  4. M0505-04Prompt templates, versioning, and testing
  5. M0505-05Getting structured JSON output from an LLM
  6. M0505-06Prompt, RAG, or fine-tune? A first decision rule
  7. M0707-01Sparse retrieval and BM25 keyword search
  8. M0707-02Dense retrieval with embeddings
  9. M0707-03Limits of embedding search: negation, recency, and authority
  10. M0707-04Vector databases and ANN indexes (HNSW, IVF)
  11. M0707-05Access control and permissions in RAG retrieval
  12. M0707-06Hybrid search: combining keyword and vector retrieval
  13. M0707-07Reranking with a cross-encoder
  14. M0707-08Assembling context: chunk order and the lost-in-the-middle problem
  15. M0707-09The complete RAG pipeline, stage by stage
  16. M0707-10Debugging RAG: retrieval failure versus generation failure
  17. M0707-11Grounding, citations, and letting a model say I don't know
  18. M0707-12When RAG is the wrong tool
  19. M0808-01How to curate a dataset for an LLM task
  20. M0808-02Data quality problems: label noise, leakage, imbalance, and drift
  21. M0808-03Exploratory data analysis (EDA) on a text corpus
  22. M0808-04Choosing the right chart: histogram, box plot, scatter, bar, heatmap, line
  23. M0808-05NVIDIA RAPIDS: cuDF, cuML, and cuGraph for GPU data science
  24. M1313-01NVIDIA's four pillars of trustworthy AI and how to implement each
  25. M1313-02NVIDIA NeMo Guardrails and content moderation for LLM applications
  26. M1313-03Prompt injection and indirect injection through RAG
  27. M1313-04Bias in AI: sources, measurement, and mitigation
  28. M1313-05Data privacy, consent, and why model weights cannot forget
  29. M1313-06Transparency, explainability, and model cards for LLM systems
  30. M1313-07AI energy use, carbon, and efficient inference
  31. M1313-08A trustworthy AI checklist for your own LLM service
  32. M1313-09How to read an AI research paper and track new LLM trends

Part of the throughlines running across the NCA-GENL prep course.