After some learnings, I have tried to summarise my understanding into different sections, and write them in LLMLLM
[[Resources for Large Language Model]]
Prerequisite
It would be easier if we have knowledge on NLP or Embedding, I would recommend Databricks course, the course start from the NLP knowledge ti...
Top Tech News
Application Example
- kapa.ai
- mendable.ai
- FinSight Financial report analyzer
- MosaicML
- GPT Engineer
- AutoGPT
- Opus.Pro - AI Video Clipping Tool
- Using LLM to create prototype of ML tasks (e.g., text classification, text extraction, or even generate labelling data) - From Snorkel AI webinar
Resources
- LLM application full architecture
- Enterprise LLM Challenges and how to overcome them - Snorkel
- Retrieval augmented generation: Keeping LLMs relevant and current
- Open Challenge for LLM Research
- What Are Large Language Model (LLM) Agents and Autonomous Agents
- A Survey of Techniques for Maximizing LLM Performance
- 10 Ways to Improve the Performance of Retrieval Augmented Generation Systems
- Pattern for LLM
- A Guide on 12 Tuning Strategies for Production-Ready RAG Applications
- # Evaluating RAG Applications with RAGAs
- RAG vs Finetuning by Heiko HotzHeiko Hotz
Principal Solutions Architect for Generative AI at AWS ◆ Founder of NLP London
Linkedin
- ACL 2023: Retrieval-based LM
- Challenges and Applications of Large Language Models
- LLMOps - Databrick
- Don't Build AI Products The Way Everyone Else Is Doing It - Builder.io
- Cohere Reranking
- HyDE
Research Paper
- A Study on Robustness and Reliability of Large Language Model Code GenerationA Study on Robustness and Reliability of Large Language Model Code Generation
Reference: https://arxiv.org/pdf/2308.10335.pdf
In short:Currently, software engineers should use tools like Co-pilot cautiously.
The evaluation results show that even for GPT-4, 62% of the gener...
Prompting
- ReAct Prompting
- Long context prompting for Claude 2.1
- Lost in the Middle: How Language Models Use Long Contexts
- PAL - Program-Aided Language Models
- Chain-of-Table Evolving Tables in the Reasoning Chain forTable UnderstandingChain-of-Table Evolving Tables in the Reasoning Chain forTable Understanding
Reference: https://arxiv.org/pdf/2401.04398.pdf
Summary
Working with table is hard
Generic Reasoning, won't work, mostly due to difficulty in extractions
Program-aided Reasoning, won't work...
Security
Tools/Library
- LangChain
- LangSmith
- LangFuse
- Fiddler
- Semantic Kernel from Microsoft
- Meta LLaMa
- OpenThaiGPT
- LlmaIndex
- WhyLogs
- Span Markers
- Ragas
- GPTCache
- Rivert - Open Source AI agent environment
Vector Databases
Courses
- Generative AI
- LangChain for LLM App DevelopmentLangChain for LLM App Development
Summary:
The LangChain for LLM Application Development Course, focuses on utilizing the LangChain Python/TypeScript framework to streamline the creation of Language Model (LLM) applications. The c...: Link - LangChain Chat with Your DataLangChain Chat with Your Data
Summary:
The LangChain Chat with Your Data course provides a step-by-step exploration of RAG EssentialsRAG Essentials
RAG Essentials :
How to split documents for embedding
How to Create high-quality embeddings.
How to choose vector store
How to search in order to retrieve most useful and relevance docum..., guiding users through fundamental processes. The course covers loading various document ...: Link - Functions, tools, and Agents with LangChainFunctions, tools, and Agents with LangChain
Summary
OpenAI Function Calling
Recent update from OpenAI to support Function calling
import json
# Example dummy function hard coded to return the same weather
# In production, this could be y...: Link - ChatGPT Prompt Engineering for DevelopersChatGPT Prompt Engineering for Developers
This is a cool course, which help you understand the tips, tactics to write a better prompt. Check it out ChatGPT Prompt Engineering for Developers
Prompting Principles
Principle 1: Write clear a...: Link - Quality and Safety for LLM ApplicationsQuality and Safety for LLM Applications
Whylogs help logs the data, as well as profiling and monitoring the data issue with LLM.
In the note, we will cover
Hallucination
Data Leakage
Refusals and Prompt Injections
Passive and a...: Link - Finetuning Large Language Models: Link
- Building and Evaluating Advanced RAGBuilding and Evaluating Advanced RAG
RAG Triad from Truera
RAG Triad
Answer Relevance
Context Relevance
Groundedness
Answer Relevance
from trulens_eval import Feedback
f_qa_relevance = Feedback(
provider.relevan...: Link - Vector Databases: from Embeddings to Applications: Link
- SentenceTransformer
- Building Generative AI Applications with Gradio: Link
- How Business Thinkers Can Start Building AI Plugins With Semantic Kernel
- Databricks: Large Language Models: Application through Production
- Databricks: Large Language Models: Foundation Models from the Ground Up
My Notes
- RAG EssentialsRAG Essentials
RAG Essentials :
How to split documents for embedding
How to Create high-quality embeddings.
How to choose vector store
How to search in order to retrieve most useful and relevance docum...
#llm #chatgpt