Build Your First LLM-based Application with Haystack

PyLadiesCon - December 2, 2023

Website: https://conference.pyladies.com/index.html
Abstract: https://pretalx.com/pyladiescon-2023/talk/FL8RG3/

Resources:
- Haystack: https://github.com/deepset-ai/haystack
- Demo Notebook: https://colab.research.google.com/drive/1PycIRK1F339vwD9KK6qNw0Mc9UPSgL_4#scrollTo=s69Upvz5MeEZ
- Advent of Haystack: https://haystack.deepset.ai/advent-of-haystack

Transcript

1. Bilge Yücel PyLadiesCon Build Your First LLM-based Application with Haystack
2. 01 - What is it? • 🥑 Developer Advocate at deepset • 🏗 Open source LLM Framework: Haystack • 📍 Istanbul, Turkey Bilge Yücel Developer Advocate 🥑 deepset Twitter: @bilgeycl Linkedin: Bilge Yucel GitHub: @bilgeyucel
3. Agenda 01 - Text Embeddings 02 - Vector Databases 03 - Retrieval Ret 04 - LLMs 05 - Build a Generative QA App
4. 01 Text Embeddings
5. Text Embeddings/Text Vectors to be or not to be • Manageable by computers • Different techniques: ◦ Sparse: TF-IDF, BM25... ◦ Dense: Trained models (Sentence Transformers, Cohere, OpenAI...) • Often has 768 dimensions
6. 02 Vector Databases
7. • Databases that store high-dimensional vectors • Optimized for vectors: ◦ Vector search ◦ CRUD operations ◦ Metadata filtering Vector Databases
8. 03 Retrieval
9. Retrieval Query • Getting the most relevant information to the query • Used for semantic search, question answering and more
10. 04 LLMs
11. Large Language Models (LLMs) • Big language models • Prompt → Human-like output • Text generation: summarization, generative QA, writing code, chat…
12. 05 Build a Generative QA Application
13. Prompting
14. LLM: Limitations • LLMs do not know the answer to everything • But they are good at following instructions • We can help them in their task by giving them the relevant context + instruction
15. Prompting
16. Prompting
17. Retrieval Augmented Generation (RAG)
18. Retrieval Augmentation Use Cases Prompt Given the following context, answer the question. If the answer is not contained within the context, say ‘I don’t know’. Question Answering Context: {{context}} Question: {{question}} Answer: Prompt Summarize the following text. Summarization Text: {{text}} Summary: Question Generation Document: {{document}} Questions: Prompt Given the following document, generate some questions
19. • Fully open-source framework built in Python for custom LLM applications • Provides tools that developers need to build state-of-the-art NLP systems • Building blocks: Pipelines & Components Haystack
20. Indexing Pipeline urls
21. Indexing Pipeline Notebook
22. Indexing Pipeline Notebook
23. Generative QA Pipeline (RAG) What is happening at OpenAI?
24. Generative QA Pipeline (RAG) Notebook
25. Generative QA Pipeline (RAG) Notebook
26. https://haystack.deepset.ai/advent-of-haystack
27. Resources Join 👇 Advent of Haystack Check out 👇 Haystack @bilgeycl Bilge Yücel Find 👇 Presentation
28. Thank you!