Introduction
Understanding grounding techniques for LLMs
()
Setting up your LLM environment
()
1. Basic LLM Hallucinations
What is a hallucination?
()
Hallucination examples
()
Comparing hallucinations across LLMs
()
Dangers of hallucinations
()
Challenge: Finding a hallucination
()
Solution: Finding a hallucination
()
2. Types of Hallucinations
Training LLMs on time-sensitive data
()
Poorly curated training data
()
Faithfulness and context
()
Ambiguous responses
()
Incorrect output structure
()
Declining to respond
()
Fine-tuning hallucinations
()
LLM sampling techniques and adjustments
()
Bad citations
()
Incomplete information extraction
()
3. Mitigating Hallucinations
Few-shot learning
()
Chain of thought reasoning
()
Structured templates
()
Retrieval-augmented generation
()
Updating LLM model versions
()
Model fine-tuning for mitigating hallucinations
()
Orchestrating workflows through model routing
()
Challenge: Automating ecommerce reviews with LLMs
()
Solution: Automating ecommerce reviews with LLMs
()
4. Detecting Hallucinations
Creating LLM evaluation pipelines
()
LLM self-assessment pipelines
()
Human-in-the-loop systems
()
Specialized models for hallucination detection
()
Building an evaluation dataset
()
Optimizing prompts with DSPY
()
Optimizing hallucination detections with DSPY
()
Real-world LLM user testing
()
Challenge: A more well-rounded AI trivia agent
()
Solution: A more well-rounded AI trivia agent
()
5. Hallucination Paper Review
Ragas: Evaluation paper
()
Hallucinations in large multilingual translation models
()
Do LLMs know what they don’t know?
()
Set the Clock: LLM temporal fine-tuning
()
Review of hallucination papers
()
Conclusion
Continue your practice of grounding techniques for LLMs
()