GENERATIVE AI SUMMARY

1. Loss metric: Measures how wrong a model's predictions are. Lower loss is better.
2. Cosine distance of 0: Indicates two embeddings are similar in direction.
3. RAG (Retrieval Augmented Generation): Uses external information to improve text generation.
4. String prompt templates: Can use any number of variables.
5. Retrievers in LangChain: Retrieve relevant information from knowledge bases.
6. Indexing in vector data: Maps vectors for faster searching.
7. Accuracy: Measures correct predictions out of total predictions.
8. Keyword-based search: Evaluates documents based on keyword presence and frequency.
9. Soft prompting: When there is a need to add learnable parameters to a Large Language Model (LLM) without task-specific training.
10. Greedy decoding: Selects the most probable word at each step in text generation.
11. T-Few fine-tuning: Updates only a fraction of model weights.
12. LangChain: Python library for building LLM applications.
13. Prompt templates: Use Python's str.format syntax for templating.
14. RAG Sequence model: Retrieves multiple relevant documents for each query.
15. Temperature in decoding: Influences probability distribution over vocabulary.
16. LLM in chatbot: Generates linguistic output.
17. Chain interaction with memory: Before and after chain execution. 18. Challenge with diffusion models for text: Text is not categorical.
18. Vector databases vs. relational databases: Based on distances and similarities.
19. StreamlitChatMessageHistory: Stores messages in Streamlit session state, not persisted.
20. Semantic relationships in vector databases: Crucial for LLM understanding and generation.
21. Groundedness vs. Answer Relevance: Groundedness focuses on factual correctness, Answer Relevance on query relevance.
22. Fine-tuning vs. PEFT: Fine-tuning trains the entire model, PEFT updates a small subset of parameters.
23. Fine-tuning appropriateness: When LLM doesn't perform well and prompt engineering is insufficient.