This article introduces Milvus, an open‑source, cloud‑native vector database designed for AI workloads, explains how it helps mitigate large‑model hallucinations, outlines its CVP architecture, showcases performance benchmarks, and explores diverse application scenarios and future directions for LLM‑vector database integration.
Code Llama, Meta’s specialized extension of Llama 2, is a large language model fine‑tuned on code data that can generate, complete, and debug software across multiple languages, supports up to 100 k tokens, and is freely available for research and commercial use.
MetaGPT is a meta‑programming framework that encodes standard operating procedures as prompts, enabling LLM‑driven multi‑agent systems to automatically generate software artifacts, coordinate roles, and build complex applications like a Blackjack CLI with near‑perfect task completion.
CoUnit is an open‑source Rust‑based tool that uses local semantic search and LLMs to create a virtual team interface, enabling low‑cost, offline knowledge retrieval, API discovery, and cross‑team assistance for software development teams.
This article investigates whether replacing traditional ID-based item encoders with massive LLMs such as GPT‑3 improves recommendation performance, by conducting extensive experiments on three real‑world datasets, analyzing performance limits, generality of item representations, and comparing against ID‑based and prompt‑based methods.
The guide outlines prompt‑engineering best practices for large language models, distinguishing base and instruction‑tuned LLMs, emphasizing clear, structured, step‑by‑step prompts, handling hallucinations, iterating through idea‑code‑data cycles, applying techniques to summarization, reasoning and expansion, managing token costs, and providing concrete OpenAI API examples.
This article outlines a practical enterprise architecture for integrating large language models into a sales assistant, detailing knowledge ingestion, vector embedding, task planning, tool usage, and iterative dialogue, while introducing AI Agent concepts and open‑source frameworks such as LangChain.
This article explains what LangChain is, outlines its core components such as Models, Indexes, Chains, Memory and Agents, provides practical code examples for building summarization and QA pipelines, and discusses future directions for LLM‑centric development.
This article explains the fundamentals of modern large language models, outlines the top five most powerful LLMs—including GPT‑4, Claude 2, Llama 2, Orca, and Cohere—and provides practical guidance on selecting and applying them across business and development use cases.
This article examines how large language models (LLMs) are reshaping enterprise applications by enabling natural‑language interfaces, automating workflows, and enhancing data analysis, while also detailing typical use cases, integration complexities, and practical solutions such as prompt engineering and vector‑database retrieval.
This tutorial introduces LangChain, an open‑source Python framework that provides unified model access, prompt management, memory, retrieval, and tool integration, enabling developers to quickly prototype AI‑driven applications using large language models and various external data sources.
This article examines the open‑source AutoDev project, outlining its assumptions about private LLMs, the three assistance modes it provides for IDEs, technical implementation details, integration challenges, and broader thoughts on LLMs as developer copilots.
This article details a step‑by‑step guide to deploying the Vicuna open‑source LLM on a single server, covering model preparation, environment setup, dependency installation, GPU and CUDA configuration, inference commands, performance evaluation, and attempted fine‑tuning, while sharing practical observations and results.
This article introduces LangChain, explains why it’s useful for building applications with large language models, walks through installation, API‑key setup, model and embedding selection, prompt engineering, chaining, memory, agents, and vector‑store indexing, and provides runnable Python code examples throughout.
This article reviews the evolution of the GPT family from GPT‑1 to GPT‑3, comparing encoder‑decoder architectures, explaining the shift from supervised fine‑tuning to zero‑shot and few‑shot learning, and highlighting the architectural and training innovations that enabled large‑scale language models.
TypeChat, an open‑source project from Microsoft’s TypeScript creator, demonstrates how large language models can translate free‑form user requests into strongly‑typed JSON structures, enabling natural‑language interfaces for applications such as coffee ordering while reducing the complexity of traditional intent‑recognition pipelines.
This article introduces LangChain, explains its architecture and core components, and provides step‑by‑step Python examples for chat models, embeddings, prompts, indexes, chains, memory, agents, and practical use‑cases such as QA bots, web search, summarization, and persistent vector stores.
This article explains how integrating large language models into industrial configuration tools creates AI‑driven features such as knowledge Q&A, automatic application generation, smart drawing, script and material generation, and outlines the three‑layer architecture, prompt engineering, and implementation lessons for developers.
This guide shows how to create a domain‑specific chatbot by combining large language models, prompt‑tuning, and Alibaba Cloud's Hologres real‑time vector store, covering prerequisites, environment setup, code configuration, vector data preparation, verification, and optional DingTalk integration.
The author builds an internal code knowledge base for the TDesign Vue‑Next library by scraping documentation, chunking and embedding texts with OpenAI’s ada model into a vector store, then retrieving relevant chunks for LLM answers, and enhances context continuity using a JavaScript AST interpreter, achieving up to 90 % query accuracy and a 20 % productivity boost.
The article analyzes key challenges in large‑language‑model pipelines—including the necessity of cold‑start pretraining, the pitfalls of reward‑model hacking, efficiency‑effectiveness trade‑offs, evaluation difficulties, and downstream fine‑tuning limits—offering practical insights for more reliable LLM development.
This guide explains how to create a domain‑specific intelligent Q&A system by leveraging prompt‑tuning, Volcano Engine Cloud Search vector store, and LangChain, covering model selection, data embedding, vector indexing, retrieval, and LLM integration with full code examples.
This step‑by‑step tutorial shows how to set up a GPU environment, prepare custom text data, train a tokenizer, configure and train a GPT‑2‑based large language model, test its generation, and run the entire pipeline using only 61 lines of Python code.
This article introduces LangChain, explains its modular architecture—including models, prompts, memory, chains, and agents—shows how to handle tokens and model selection, and provides complete Python code examples for building a conversational retrieval chatbot with OpenAI models.
This comprehensive guide explores prompt engineering for large language models, covering its background, fundamental concepts, prompt formats, construction principles, advanced techniques like few‑shot, zero‑shot, and chain‑of‑thought prompting, as well as practical examples, evaluation metrics, and future directions.
This article analyzes the evolving landscape of large‑model applications, detailing prompt engineering, engineering challenges, AI‑engineer roles, domain‑driven design, and agent frameworks, while offering practical guidance and references for building production‑grade LLM‑driven systems.
The article details a workflow that converts the PyTorch ChatGLM‑6B model to MNN, splits and compresses embeddings, applies int4/int8 quantization, supports dynamic shapes, and uses hybrid GPU/CPU or CPU‑only loading to enable low‑memory edge inference on PCs and mobile devices with competitive token‑per‑second performance.
This article reviews how recent large language models (LLMs) are reshaping recommendation systems, covering their emergence, in‑context learning, prompt‑based strategies, three main LLM‑driven recommendation paradigms, key research papers, experimental results, and future research directions.
This article explains the limitations of large language models, introduces prompt engineering as a remedy, and provides a comprehensive guide to using the LangChain framework—including models, prompts, chains, agents, vector search, and practical Node.js code examples—to enable LLMs to interact with external tools and data sources.
This article explores how large language models can be integrated into software architecture governance, turning architectural rules into code, enhancing design, development, and runtime phases, and improving code generation quality through explicit knowledge, DSLs, and full‑process AI assistance.
LangChain is an innovative framework that simplifies building language‑model‑driven applications by providing modular components such as models, prompts, memory, chains, and agents, with examples of asynchronous LLM calls, custom prompt templates, vector stores, token handling, and a simple Gradio chatbot implementation.
The article introduces LangChain, a framework that unifies language models, prompts, memory, retrieval, and tool‑driven agents into composable chains, illustrating its core components, code examples, end‑to‑end applications such as retrieval‑augmented QA and image generation, and outlining future uses in customer service, recommendation, and automated code review.
Microsoft's newly released Guidance language unifies generation, prompting, and logical control for large language models, enabling seamless integration with providers like Hugging Face, enhancing prompt reliability, and simplifying complex LLM workflows for developers.
Vector databases enable efficient similarity search and storage of high‑dimensional embeddings, allowing enterprises to combine large language models with proprietary knowledge assets to create domain‑specific, accurate, and up‑to‑date AI services, as illustrated with open‑source solutions Chroma and Milvus.
This article explains what vectors and vector databases are, how they differ from traditional databases, their core technologies, their relationship with large language models, market trends, and details of Tencent Cloud VectorDB’s capabilities, architecture, real‑world applications, and future competitive challenges.
Vector databases store and search large-scale vector data, and in 2023 the five leading solutions—Chroma, Pinecone, Weaviate, Milvus, and Faiss—offer scalable, high-performance options for applications such as LLM-driven services, audio search, recommendation systems, image/video analysis, and semantic retrieval across various industries.
This article explores how large language models can standardize software engineering practices by converting requirements and designs into DSL formats, enabling more automated and efficient code generation, while discussing the challenges of dynamic context building, DSL specification, and the evolving role of LLMs in development pipelines.
The article examines the rapid rise of GPT models, their technical capabilities and limitations, and how their integration is reshaping software interaction from command‑line to GUI‑plus‑Language UI, offering frontend engineers new opportunities, practical examples, and guidance on leveraging large‑model AI in product design.
This article presents a detailed reference architecture for building LLM‑driven applications, using the ArchGuard Co‑mate project to illustrate layered design, local model integration, DSL‑based orchestration, and streaming LLM interfaces, complete with code examples and practical implementation notes.
Databricks' $1.3 billion acquisition of MosaicML brings the startup's open‑source MPT models and high‑efficiency training stack into the Lakehouse platform, reflecting a strategic push to embed generative AI across enterprises while emphasizing data control, cost reduction, and open‑source policies.
The A2M Internet Architecture and AI Summit in Shanghai showcased 24 topics and 84 talks, highlighting AIGC applications across enterprise collaboration, large‑scale model integration, 3D visual content, music entertainment, and the broader impact of LLMs on software development and digital transformation.
This article outlines a reference stack for LLM applications, detailing the context‑learning design pattern, key components such as vector databases and embedding models, orchestration frameworks like LangChain, and discusses trade‑offs between proprietary and open‑source models, scaling challenges, and the future role of AI agents.
This guide walks through deploying a Bloom 7B1 large language model for distributed inference on Alibaba Cloud Container Service (ACK) using DeepSpeed, Arena, and Kubernetes, covering environment setup, model configuration, service launch, verification, and Ingress exposure.
DeepKE-LLM is an open‑source, extensible knowledge‑graph extraction framework that leverages large language models for entity, relation, and attribute extraction, supports multiple LLMs, provides installation scripts, various usage modes, fine‑tuning pipelines, and integrates with the KnowLM project for advanced instruction‑following capabilities.
This guide explains how to deploy large language models on Alibaba Cloud's ACK using KServe for serverless inference, integrates Fluid for distributed data caching to cut cold‑start latency, provides step‑by‑step commands, performance benchmarks, and practical tips for production‑grade AI model serving.
A NYU Tandon study shows that GPT‑4, using plain‑language dialogue, can generate Verilog code for a novel 8‑bit accumulator microprocessor, which was tape‑out on Skywater 130 nm, demonstrating AI’s emerging ability to perform hardware design tasks at near‑human level.
The announcement introduces RPI's prestigious computer science department, its extensive GPU resources, collaborations with IBM Research, and detailed profiles of three incoming faculty members—highlighting their research areas in graph neural networks, trustworthy AI, data‑centric AI, drug‑design generative models, and neural‑symbolic reasoning—while inviting PhD and intern applicants to apply with full scholarships and funding support.
The article outlines five practical patterns—natural‑language DSL, streaming DSL, DSL‑guided generation, explicit retry, and dynamic proxying—that enable developers to treat large‑language‑model interactions as first‑class APIs, improving efficiency, accuracy, and user experience across diverse scenarios.
A comprehensive look at how companies are rapidly adopting large language model APIs, retrieval techniques, and custom model strategies, revealing key statistics, emerging toolchains, and the shifting balance between closed‑source LLM services and open‑source custom stacks.
This article explains how to craft efficient prompts by combining clear instructions and questions, discusses prompt injection risks and mitigation with delimiters, addresses hallucinations, and introduces zero‑shot, few‑shot, and chain‑of‑thought prompting techniques for large language models.
This article outlines four foundational design principles—user‑intent‑driven design, context awareness, atomic capability mapping, and language‑API integration—for building LLM‑centric software architectures, illustrating each with DSL examples, Kotlin implementations, and practical insights on prompt engineering, dynamic context layering, and API evolution.
This article explains how ArchGuard 2.0 leverages large language models to transform tacit architectural knowledge into dynamic, DSL‑based governance functions, detailing the challenges of traditional tools, the design of the Co‑mate system, and the practical implementation using Kotlin REPL.
This article explains what embeddings and vector databases are, why they matter for large language models, and how they can be used to overcome token limits by storing and retrieving relevant text chunks for accurate, context‑aware responses.
The article explains prompt‑engineering fundamentals—definitions, instruction, context, and output formatting—and showcases tricks like few‑shot, chain‑of‑thought, and ReAct, then demonstrates testing with OpenAI APIs, token management, LangChain integration, and low‑code applications such as AI‑generated SQL, API gateways, DSL‑driven UI, chart creation, and vector‑based semantic search.
This comprehensive guide walks programmers through the fundamentals of large language model capabilities, prompt writing and management, new interaction and workflow designs, advanced scenario‑specific applications, and context engineering, offering practical strategies and architectural insights for AI‑native development.
LangChain, launched in 2022, quickly evolved from a Python library to a multi‑environment framework that helps developers build chat‑based applications, agents, and memory‑aware LLM solutions, while integrating with major cloud and AI tooling ecosystems.
This article explains what embeddings and vector databases are, how they are generated with models like OpenAI's Ada, why they enable semantic search and help overcome large language model token limits, and demonstrates a practical workflow for retrieving relevant document chunks using cosine similarity.
The DataFunCon2023 Digital Humans and XR Interaction Forum showcases the latest advances in digital human technology, multimodal interaction, emotion computing, and LLM‑enhanced AI beings through expert talks, detailed outlines, and audience benefits, highlighting both research progress and practical applications.
Microsoft researchers fine‑tuned GPT‑3.x models with LoRA on over 40,000 cloud incident records, evaluated them with six NLP metrics and human interviews, and found that LLMs can generate root‑cause analyses and mitigation steps comparable to BERT models, especially for machine‑detected failures.
This guide details the end‑to‑end process of installing GPU drivers, setting up a Python environment, deploying the open‑source Alpaca‑LoRA model, fine‑tuning it with Chinese data on a multi‑GPU server, and performing inference, while highlighting practical challenges and performance observations.
This article explains why combining large language models with vector databases enables enterprises to create specialized, up‑to‑date chatbots, outlines the underlying principles, describes the ADB‑PG vector‑search capabilities, and provides step‑by‑step implementation details including data processing, indexing, and query examples.
This article reverse‑engineers GitHub Copilot's context construction, detailing visible and invisible processes, prompt composition, token allocation strategies, fast response mechanisms, and future directions for LLM context engineering.
This article analyzes the current landscape of large‑language‑model application frameworks—LangChain, Auto‑GPT, and HuggingGPT—detailing their core capabilities, workflow patterns, limitations, and how they enable developers to integrate LLMs into real‑world products while outlining future platform requirements.
A recent study by the LMSYS organization introduces an Elo‑rated, 1v1 battle arena for large language models, ranking open‑source chatbots like Vicuna, Koala, and ChatGLM, while discussing the limitations of traditional benchmarks and the advantages of crowd‑sourced, scalable evaluation.
This article explains why and how to fine‑tune the LLaMA‑7B model using the cheap Alpaca‑LoRA approach, covering hardware requirements, dataset preparation, LoRA training, optional model merging and quantization, and provides ready‑to‑run code snippets for single‑ and multi‑GPU setups.
AutoGPT, the latest autonomous AI system, can independently browse the web, use third‑party tools, and execute tasks without human intervention, exemplified by building a React website in minutes, and this article explains its principles, showcases demos, and provides step‑by‑step installation instructions.
This article examines Auto‑GPT’s hype versus reality, detailing its architecture, cost per task, looping pitfalls, limited memory management, and why its current capabilities fall short of practical production use.
The article surveys major large language models—including GPT‑3, T5, LaMDA, Jurassic‑1, MT‑NLG, Gopher, Chinchilla, PaLM, U‑PaLM, OPT, LLaMA, BLOOM, GLM‑130B, and ERNIE 3.0 Titan—explains their architectures, scaling trade‑offs, and then details instruction‑fine‑tuned variants such as T0, FLAN, GPT‑3.5, ChatGPT, GPT‑4, Alpaca and ChatGLM, providing references for further study.
The Vicuna-13B open‑source chatbot, fine‑tuned from LLaMA on 70k ShareGPT conversations, matches over 90% of ChatGPT and Google Bard quality while costing only about $300 to train, thanks to memory optimizations, multi‑turn dialogue handling, and cheap spot‑instance training.
In a SETalk livestream, experts discuss how ChatGPT and Copilot X dramatically boost coding efficiency, lower entry barriers, reshape software development workflows, raise concerns about job displacement, and outline three emerging paradigms—LLM for SE, SE for LLM, and LLM as SE—while highlighting current limitations and future trends.
This article reviews the rapid progress of large language models (LLMs), covering their historical development, scaling laws, emergent abilities, core technologies such as training and alignment, resource ecosystems, evaluation methods, safety concerns, and prospective research challenges.
A collection of Zhihu answers reflects on how the release of GPT‑4 has reshaped NLP research, dividing the community into LLM‑enthusiasts and skeptics, discussing the relevance of parsing, resource‑driven research directions, and the existential challenges faced by researchers.
The article examines ChatGPT's rapid adoption, the rise of generative AI, and how integrating external knowledge sources such as knowledge graphs can improve factual accuracy, enhance downstream tasks, and enable a variety of ChatGPT+ applications ranging from intelligent search to office automation tools.
Microsoft's open‑source Visual ChatGPT extends ChatGPT with image send/receive capabilities, explains its multimodal architecture, demo scenarios, used visual models, and points to the arXiv paper, highlighting its rapid popularity growth on GitHub.
Prompt‑Ops frameworks like LangChain let developers turn pre‑trained LLMs into versatile applications by abstracting model calls, chaining prompts, integrating tools, managing memory, and handling private data, while addressing challenges such as nondeterminism, version control, and prompt injection in production environments.
This report investigates the impact of large language model‑based AI coding assistants such as Copilot, Tabnine, CodeGeeX, and ChatGPT on frontend and backend development, evaluates their efficiency gains, compares features, pricing, and IDE integration, and provides practical adoption recommendations for development teams.
This article analyses the economic feasibility of integrating large language models (LLMs) into search, estimating inference and training costs, exploring hardware efficiency, scaling laws, and future trends, and concludes that while technically viable, the added expense may challenge profitability for major search providers.
This article examines whether large language model‑driven search can replace traditional engines like Google, analyzing the economic feasibility, training and inference costs, hardware considerations, and future scaling trends, while providing a detailed cost framework and insights into the profitability of LLM‑based services.
The article recounts a hands‑on exploration of ChatGPT’s abilities for database administration, highlighting impressive language understanding, notable mistakes such as confusing pg_resetwal with pg_resetxlog, and the broader implications of large language models for AI‑driven search and workflow automation.
This article surveys recent open‑source projects—including LAION’s OpenAssistant, CarperAI’s trlX, and Phil Wang’s ChatGPT implementation—that provide RLHF‑based training pipelines for large language models, while highlighting community expectations, resource challenges, and future accessibility goals.
The article reviews the evolution of large language models from the pre‑GPT‑3 era to the present, analyzes the conceptual and technical gaps between Chinese and global research, and outlines key future research directions such as scaling laws, prompting techniques, multimodal training, and efficient model architectures.
The article explains ChatGPT’s underlying technology, detailing its three-stage training pipeline—supervised fine‑tuning, reward‑model learning, and reinforcement learning with PPO—while discussing its strengths, limitations, and potential integration with traditional search engines.
Prof. Qiu Xipeng’s talk highlighted how large language models can be offered as a service and efficiently adapted via in‑context learning, lightweight label‑tuning, and gradient‑free black‑box optimization, showcasing a unified asymmetric Transformer (CPT) that handles understanding, generation, ABSA and NER tasks while reducing resource demands.
