Content

From company news to technical tutorials – explore the most popular content on the Zilliz blog.

Vector Lakebase Stories

Seamless AI Workflows

Start Free, Scale Easily

Start to build your GenAl apps today with Zilliz Cloud Serverless

AI Assistant

Ruben Winastwan

Discover how Chain-of-Agents enhances Large Language Models by effectively managing context injection, improving response quality while addressing token limitations.


Chain of Agents (COA): Large Language Models Collaborating on Long-Context Tasks

NVIDIA's Cosmos platform enables safe, digital twin training of GenAI models for physical applications, overcoming data scarcity and safety challenges.

Cosmos World Foundation Model Platform for Physical AI

Learn how to build Retrieval Augmented Generation (RAG) applications using Dify for orchestration and Milvus for vector storage in this step-by-step guide.

Building RAG with Dify and Milvus

These multimodal RAG patterns include grounding all modalities into a primary modality, embedding them into a unified vector space, or employing hybrid retrieval with raw data access. 


3 Key Patterns to Building Multimodal RAG: A Comprehensive Guide 

RA-DIT, or Retrieval-Augmented Dual Instruction Tuning, is a method for fine-tuning both the LLM and the retriever in a RAG setup to enhance overall response quality.

Enhancing RAG with RA-DIT: A Fine-Tuning Approach to Minimize LLM Hallucinations

We discussed the checkability training approach to help LLMs generate accurate answers that humans can easily understand and verify. 


Prover-Verifier Games Improve Legibility of LLM Outputs

RouteLLM is an open-source framework that enables developers to efficiently route tasks to the most suitable LLMs based on cost, latency, and accuracy.

RouteLLM: An Open-Source Framework for Navigating Cost-Quality Trade-Offs in LLM Deployment

LoRA (Low-Rank Adaptation) is a technique for efficiently fine-tuning LLMs by introducing low-rank trainable weight matrices into specific model layers.

LoRA Explained: Low-Rank Adaptation for Fine-Tuning LLMs

Robert Tran, the Co-founder and CTO of Inkeep, shared how Inkeep and Zilliz built an AI-powered assistant for their documentation site. 

How Inkeep and Milvus Built a RAG-driven AI Assistant for Smarter Interaction

HNSWlib is an open-source C++ and Python library implementation of the HNSW algorithm, which is used for fast approximate nearest neighbor search. 

Understanding HNSWlib: A Graph-based Library for Fast Approximate Nearest Neighbor Search

In this article, we explored various RAG components and discussed the approaches with optimal performance in each component. 

Best Practices in Implementing Retrieval-Augmented Generation (RAG) Applications 

In this article, we'll explore the evolution of MAS from a methodological or approach-based perspective.

The Evolution of Multi-Agent Systems: From Early Neural Networks to Modern Distributed Learning (Methodological)

In this article, we'll discuss the evolution of MAS from its early days to the most recent developments from an algorithmic perspective. 

The Evolution of Multi-Agent Systems: From Early Neural Networks to Modern Distributed Learning (Algorithmic) 

PagedAttention and vLLM solve important challenges in serving LLMs, particularly the high costs and inefficiencies in GPU memory usage when using it for inference. 


Efficient Memory Management for Large Language Model Serving with PagedAttention

Deep residual learning solves the degradation problem, allowing us to train a neural network while still potentially improving its performance.

Deep Residual Learning for Image Recognition

Zilliz Cloud Serverless allows users to store, index, and query massive amounts of vectors at only a fraction of the cost while keeping a competitive level of performance.  


Up to 50x Cost Savings for Building GenAI Apps Using Zilliz Cloud Serverless

GPU-accelerated vector search through NVIDIA's cuVS library and CAGRA algorithm are highly beneficial for optimizing AI app performance in production. 

Milvus on GPUs with NVIDIA RAPIDS cuVS 

We explore the combination of time series databases and vector databases, focusing on the use case in Internet of Things (IoT) applications.

Improving Analytics with Time Series and Vector Databases

Boolean retrieval models are an information retrieval method for retrieving relevant documents using a set of Boolean logic operators, eg. AND, OR, and NOT.


Understanding Boolean Retrieval Models in Information Retrieval 

Constitutional AI (CAI) is a concept in artificial intelligence focused on ensuring that AI systems operate in alignment with ethical and legal principles


Constitutional AI: Harmlessness from AI Feedback

Multimodal RAG extends RAG by accepting data from different modalities as context. Learn how to build one with Gemini 1.5, BGE-M3, Milvus, and LangChain. 


Building a Multimodal RAG with Gemini 1.5, BGE-M3, Milvus Lite, and LangChain 

LLM-Eval is an approach to simplifying and automating the evaluation of LLM conversation quality.


LLM-Eval: A Streamlined Approach to Evaluating LLM Conversations 

Retrieval Augmented Generation (RAG) is a method for mitigating LLM hallucinations. Learn how to build a chatbot RAG with Milvus, Llama3, and LlamaIndex. 

Building RAG with Milvus Lite, Llama3, and LlamaIndex

Regularization prevents a machine-learning model from overfitting during the training process. We'll discuss its concept and key regularization techniques. 

Understanding Regularization in Neural Networks 

A step-by-step process to build an efficient and production-ready vector search pipeline using Databricks Spark and Milvus. 

Building Production Ready Search Pipelines with Spark and Milvus

With Snowflake's Snowpark Container Service (SPCS), users can now run Milvus within the Snowflake ecosystem, allowing them to easily interact with Milvus using data stored in Snowflake.

Building RAG with Self-Deployed Milvus Vector Database and Snowpark Container Services

This article discusses several options for customizing LLMs to enhance their performance on specific tasks. 

Introduction to LLM Customization

This article discussed how Delivery Hero used AI models to generate high-quality food images to improve user experience and conversion rate. 

How Delivery Hero Implemented the Safety System for AI-Generated Images

This post explains how to extract content from a website and use it as context for LLMs in a RAG application. However, before doing so, we need to understand website fundamentals.

A Beginner's Guide to Website Chunking and Embedding for Your RAG Applications

In this post, we'll explore several vector indexing strategies that can be used to efficiently perform similarity search, even in scenarios where we have large amounts of data and multiple constraints to consider.


How to Pick a Vector Index in Your Milvus Instance: A Visual Guide

In general, there are two types of vectors: dense vectors and sparse vectors. While they can be utilized for similar tasks, each has advantages and disadvantages. In this post, we will delve into two popular variants of sparse embedding: BM25 and SPLADE. 


Comparing SPLADE Sparse Vectors with BM25

You will learn how Zilliz Cloud Pipeline transforms PDF data into a format ready for LLMs to use in semantic search tasks. Finally, we will conduct data retrieval using vector search.


Transforming PDFs into Insights: Vectorizing and Ingesting with Zilliz Cloud Pipelines

If we store all of the data, including the chunk text and the embedding, inside of Pandas DataFrame, we can easily integrate and import them into the Milvus vector database.

Pandas DataFrame: Chunking and Vectorizing with Milvus

Explore the concepts of vector embeddings and vector databases and their role in improving the user experience in e-commerce. 


Leveraging Vector Databases for Next-Level E-Commerce Personalization

Learn about time series data including general concepts and preprocessing methods to transform time series data into an embedding suitable for forecasting tasks.

Time Series Embedding in Data Analysis

Delve into one of the deep learning models that has played a significant role in the development of sentence embedding: MPNet. 


All-Mpnet-Base-V2: Enhancing Sentence Embedding with AI

In this article, we'll discuss how having too much data can hinder the performance of our machine-learning model and what we can do to address this problem.


Streamlining Data: Effective Strategies for Reducing Dimensionality

Everything you need to know about the Transformers model, exploring its architecture, implementation, and limitations

Transforming Text: The Rise of Sentence Transformers in NLP

Class Activation Mapping (CAM) is used to visualize and understand the decision-making of convolutional neural networks (CNNs) for computer vision tasks. 

Class Activation Mapping (CAM): Better Interpretability in Deep Learning Models

We can easily implement the BM25 algorithm to turn a document and a query into a sparse vector with Milvus. Then, these sparse vectors can be used for vector search to find the most relevant documents according to a specific query.
