AI knowledge Page 5

Retrieval Augmented Generation (RAG) is becoming one of the most popular applications for Large Language Models (LLMs) and vector databases.RAG is the process of augmenting the input to a LLM with context retrieved from vector databases (e.g., Weaviate).The RAG application passes...

A Multi-Agent System (MAS) is a computing system consisting of multiple interacting Intelligent Agents. Multi-Agent Systems can be used to solve problems that are difficult or impossible to solve by a single Intelligent Agent or a single system. Intelligent agents can be robots, humans, or soft...

First, LLMs already have strong capabilities, why do we still need RAG (Retrieval Augmented Generation)? Although LLMs have demonstrated significant capabilities, the following challenges still warrant attention: Illusion problem: LLMs use a statistically based probabilistic approach to generate text word by word, a mechanism that inherently leads to the possibility of...

o3 is here to share some personal insights. Progress on Test-time Scaling Law has been much faster than we thought. But I'd like to say that the path is actually a bit convoluted - it's OpenAI's way of saving the country from the curve in its pursuit of AGI. Reinforcement Learning and Shortcut Thinking For ...

Vector Embedding is the core of current Retrieval Augmented Generation (RAG) applications. They capture semantic information of data objects (e.g., text, images, etc.) and represent them as arrays of numbers. In current generative AI applications, these vector Embedding are usually generated by Embedding models. How to apply for RAG ...

PREFACE Over the past two years, Retrieval-Augmented Generation (RAG, Retrieval-Augmented Generation) technology has gradually become a core component for enhancing intelligences. By combining the dual capabilities of retrieval and generation, RAG is able to introduce external knowledge, thus providing more applications of large models in complex scenarios...

Agent The most common translation I've seen so far is "intelligent body", but the direct translation is "agent". What does Agentic translate to? I feel that a word like "agentic" is more appropriate. So in order not to confuse the readers, I use English directly in this article. With the development of LLM, the ability of AI...

In order for an AI model to be useful in a particular scenario, it usually needs access to background knowledge. For example, a customer support chatbot needs to understand the specific business it serves, while a legal analysis bot needs to have access to a large number of past cases. Developers often use Retrieval-Augmente...

Full Process of Fine-tuning Large Models It is recommended to strictly follow the above process during fine-tuning and avoid skipping steps, which may lead to ineffective labor. For example, if the dataset is not fully constructed, and it is eventually found that the poor effect of the fine-tuned model is a problem of the quality of the dataset, then the preliminary efforts will be wasted, and the matter...

OlaChat AI Digital Intelligence Assistant 10,000-word in-depth analysis to bring you to the past and present of Text-to-SQL technology. Thesis: Next-Generation Database Interfaces: a Survey of LLM-based Text-to-SQL Generating accurate SQL from natural language problems (text-to-SQL) is a long...

01.Background In RAG application development, the first step is to chunk the document, efficient document chunking can effectively improve the accuracy of the subsequent recall content. Efficient document chunking can effectively improve the accuracy of the subsequent recalled content. How to efficiently chunk is a hot topic of discussion, there are such as fixed-size chunking, random-size chunking, sliding window...

Over the past year, we've worked with teams building Large Language Model (LLM) agents across multiple industries. Consistently, we have found that the most successful implementations did not use complex frameworks or specialized libraries, but rather were built with simple, composable patterns. In this post, we'll share our experience working with customers and since...

AI Summary Overview An in-depth look at AI cue engineering, with a roundtable format in which several experts from Anthropic share their understanding and practical experience of cue engineering from a variety of perspectives, including research, consumer, and enterprise. The article details the definition of cue engineering, its importance, and how...

Scaling Test-Time Compute has become one of the hottest topics in AI circles since OpenAI released the o1 model. Simply put, instead of piling up computational power in the pre-training or post-training phases, it is better to do it in the inference phase (i.e., when the large language model generates the output...

  Looking back to 2024, the big models are changing day by day, and hundreds of intelligent bodies are competing. As an important part of AI applications, RAG is also a "swarm of heroes and lords". At the beginning of the year ModularRAG continued to heat up, GraphRAG shine, open source tools in full swing in the middle of the year, the knowledge graph re-innovation opportunity, the end of the year graphical reasoning ...

In recent years, with the rapid development of Generative AI (GAI) and Large Language Model (LLM), their security and reliability issues have attracted much attention. A recent study has discovered a simple but efficient attack method called Best-of-N jailbreak (BoN for short). By inputting ...

Long Text Vector Modeling The ability to encode ten pages of text into a single vector sounds powerful, but is it really practical? Many people think... Not necessarily. Is it okay to use it directly? Should it be chunked? How to divide the most efficient? This article will take you in-depth discussion of different chunking strategies for long text vector models, analyzing the pros and cons...

  I. The Root Cause of Testing Prompts: LLM is highly sensitive to prompts, and subtle changes in wording can lead to significantly different outputs Untested prompts can produce: Factually incorrect information Irrelevant replies Unnecessary wasted API costs II. Systematic Optimization of Prompts ...