rag

The official companion code repository for O'Reilly's bestselling book 'Hands-On Large Language Models', featuring nearly 300 illustrations and Jupyter Notebooks that run directly on Google Colab, covering full-stack practices including Transformers, prompt engineering, RAG, multimodal models, and fine-tuning. Has earned 30K+ Stars.

This paper introduces Aligned RAG Evaluation, a matched-query protocol showing that attack success rate (ASR) hides defense-induced failures in RAG knowledge-poisoning defenses, distinguishing prevented attacks from newly introduced target-positive answers on HotpotQA and Natural Questions.

该论文研究了RAG系统中的从众偏差：当检索到的多数文档是互相一致的投毒文档时，Gemma-3-12b检测中毒声明的能力从25%显著下降至10%，表明语言模型会随附多数证据而忽视事实核查。

A preprint tests five agent memory systems (Graphiti, Zep, mem0, langmem, cognee) for policy revocation, finding no system reliably prevents agents from retrieving and acting on superseded harmful policies, with revoked records sometimes ranked first and unsafe actions taken in over 40% of trials.

An educational walkthrough of what happens between a user's message and a chatbot's reply, covering tokenization, context windows, conversation memory, chunking/embeddings, RAG retrieval, system prompts, function calling, and safety evaluation.

A practitioner shares their struggle balancing free-form LLM writing with strict factual traceability in a long-form generation system, detailing failed template and structured-citation approaches and asking the community whether a post-hoc claim extraction and verification architecture works better in production.

I recommend Manning’s new book, *Domain-Specific Small Language Models: Efficient AI for local deployment* (by Guglielmo Iozzia, published in May 2026), which explains how to fine-tune small models on domain-specific data, quantize them, and deploy them on local hardware to achieve high-quality performance on specialized tasks. It also covers RAG and agent systems that rely solely on small models.

DAIR.AI发布了本周（10月26日至11月1日）热门AI论文精选，涵盖HarnessMM、DecisionRAG、Unified Training、Agentic Backdoors、EmbeddingGemma 2等研究方向。

Free open-source 'AI Engineering from Scratch' curriculum teaches building backprop, tokenizers, attention, transformers, hybrid RAG, and agents from raw math across Python, TypeScript, Rust, and Julia, with interactive in-terminal tutoring and runnable simulations.

A developer asks how to make local-model research and search (using Crawl4AI + SearXNG) perform on par with ChatGPT/Codex, reporting missing, outdated, and low-relevance results with models like GLM 5.3 Flash and DeepSeek V4.1 Flash across harnesses such as ZCode and OpenCode.

LangChain Community Spotlight featuring Kacper Klusek of Soba Labs on building enterprise RAG by fixing retrieval first: contextual chunking, BM25 hybrid search, and a reranker cut failed retrievals by 67% in Anthropic's evaluation, with LangSmith evals catching document drift.

A Chinese-language self-study tutorial for Stanford’s CS329Z “Engineering AI Agents” course has been released, featuring 19 lectures and 4 hands-on exercises covering LLMs, RAG, tool calling, memory, multi-agent systems, and evaluation, with support for offline HTML reading.

介绍康奈尔大学与 ASAPP 的 DecisionRAG 论文：用决策模型（如 Jev 1.13）在文档树上导航取代向量相似度检索，实现 retriever-free RAG，在 Natural Questions 上 top-1 命中率达 65.4%，优于 embedding 检索加 8B reranker 的 48.0% 和 GPT-5 reranking 的 60.8%。

A team from Zhejiang University has open-sourced the Chinese-language textbook “Foundations of LLMs,” which has earned 16K+ stars on GitHub. The complete six-chapter PDF is available for free download, covering language model fundamentals, LLM architectures, prompt engineering, parameter-efficient fine-tuning, model editing, and RAG, with ongoing monthly updates.

FilingLoop introduces an interactive benchmark and self-evolving LLM agent framework for corporate disclosure review and revision, grounded in 380 SEC IPO review tasks with requirement-level evaluation that improves fulfillment from 24.28% to 27.96% via procedural learning without weight updates.

The GitHub repository cs329z-chinese organizes the publicly available materials from Stanford’s CS329Z “Engineering AI Agents” course into a Chinese-language self-study tutorial, featuring 19 lectures, 4 hands-on exercises, and a web version that can be read offline. It is designed to help developers who know Python systematically learn agent development.

有人开源了被称为"AI知识图谱版Terraform"的工具 Open Ontologies，可让AI推理变得可验证，支持增量修改预览、事实校验与RAG声明验证，并通过MCP集成Claude Code、Cursor、Windsurf等工具。

A practitioner shares scores for AI agent retrieval/RAG tools from a rankings site they built, noting LlamaIndex leads at 7.0 while Pinecone and Ragas lag on release freshness, and asks production teams about their retrieval stacks and evaluation practices.

A guide on building a research-oriented AI agent that takes a research-style question, plans, searches the web, filters noise, and returns a structured, source-backed answer with confidence scoring and uncertainty disclosure.

A comparison of seven popular LLM evaluation tools, split between in-code libraries (promptfoo, DeepEval, Ragas) and tracing/experiment platforms (Langfuse, Arize Phoenix, LangSmith, Braintrust), with guidance on when to use each.