Retrieval Augmented Generation

via “retrieval-augmented generation (rag) pipeline composition”

Typescript bindings for langchain

Unique: RetrievalQA is a pre-built chain that combines a Retriever (vector store query interface) with a PromptTemplate and LLM. The chain automatically formats retrieved documents into context and passes them to the LLM. Multiple retrieval strategies (similarity, MMR) are supported through the Retriever interface, enabling optimization for different use cases.

vs others: More accessible than building custom RAG pipelines because it provides a standard pattern, and more flexible than monolithic RAG frameworks because retrievers, prompts, and LLMs are swappable.

via “retrieval-augmented generation (rag) with configurable engines and semantic search”

Multi-agent software company simulator — PM, architect, engineer roles collaborate on projects.

Unique: Implements a pluggable RAG system with support for multiple retrieval engines (vector, BM25, hybrid) and per-role/per-action configuration. RAG is integrated into the action framework, enabling agents to automatically augment prompts with retrieved context before LLM invocation.

vs others: More flexible than single-engine RAG systems because it supports multiple retrieval strategies and allows fine-grained configuration per role/action. Compared to external RAG pipelines, MetaGPT's RAG is tightly integrated with the agent framework and automatically handles context injection.

via “retrieval-augmented agent with memory and knowledge integration”

Microsoft AutoGen multi-agent conversation samples.

Unique: Memory systems are decoupled from agent logic via autogen-ext, allowing agents to work with any memory backend (vector DB, knowledge graph, custom) without modifying agent code; supports both pre-retrieval (before agent turn) and post-generation (refining responses) RAG patterns

vs others: More modular than LangChain's RAG chains because memory backends are truly pluggable and agents don't depend on specific vector store implementations

via “retrieval-augmented generation (rag) with long context understanding”

Databricks' 132B MoE model with fine-grained expert routing.

Unique: Leading RAG performance among open models through 32K context window, instruction-tuning for information synthesis, and fine-grained MoE routing that maintains coherence across dense retrieved context; native integration with Databricks Vector Search ecosystem

vs others: Competitive with GPT-3.5 Turbo on RAG tasks while being open-source and self-hostable; 32K context enables single-pass RAG without iterative retrieval for most document sets; more efficient than dense models due to MoE architecture

via “retrieval-augmented generation (rag) with vector stores and document readers”

Build and run agents you can see, understand and trust.

Unique: Integrates RAG through a Knowledge Base abstraction that works with pluggable vector stores and document readers, allowing agents to augment reasoning with retrieved context while maintaining separation between retrieval logic and agent reasoning

vs others: More modular than LangChain's RAG because vector stores and document readers are pluggable; more integrated than AutoGen's RAG support because it's built into the agent framework rather than requiring external libraries

via “retrieval-augmented-generation-with-vector-search”

Sample code and notebooks for Generative AI on Google Cloud, with Gemini Enterprise Agent Platform

Unique: Vertex AI's RAG Engine provides managed corpus lifecycle (ingestion, chunking, embedding, indexing) without requiring separate vector database infrastructure. The implementation uses Vector Search 2.0's streaming index updates and automatic sharding for sub-millisecond retrieval at scale, integrated directly into Gemini's context management layer.

vs others: Eliminates the need to manage separate vector databases (Pinecone, Weaviate) by providing end-to-end RAG as a managed service, and offers better cost efficiency than self-hosted solutions because embedding generation and retrieval are co-located in the same GCP region.

via “retrieval augmented generation system design and implementation”

A one stop repository for generative AI research updates, interview resources, notebooks and much more!

Unique: Organizes RAG design around explicit decision points (retriever type, embedding model, vector database, ranking strategy) with research-backed guidance on trade-offs. Includes dedicated section on agentic RAG patterns for knowledge-grounded agent decision making.

vs others: More comprehensive than framework-specific RAG documentation; provides cross-framework architectural patterns and research-backed design guidance, whereas most RAG resources focus on implementation in a specific framework.

via “retrieval-augmented generation (rag) embedding support with vector database integration”

sentence-similarity model by undefined. 17,78,169 downloads.

Unique: Embeddings are trained with a focus on retrieval tasks (MTEB retrieval benchmark), optimizing for high recall and ranking quality. The model achieves strong performance on NDCG@10 metrics, indicating effective ranking of relevant documents, which is critical for RAG quality.

vs others: Specifically optimized for retrieval tasks unlike general-purpose embeddings, and compatible with all major RAG frameworks (LangChain, LlamaIndex) through standardized vector database integration.

via “retrieval-augmented generation with embedding-based knowledge retrieval”

Agent S: an open agentic framework that uses computers like a human

Unique: Integrates RAG with procedural memory through embedding-based retrieval, enabling dynamic knowledge selection based on task context without explicit prompt engineering or context window constraints

vs others: Provides more flexible knowledge integration than static prompts while being more scalable than in-context learning with large knowledge bases

via “rag (retrieval-augmented generation) system implementation”

📚 从零开始构建大模型

Unique: Implements RAG as a modular pipeline with separate, swappable components for embedding generation, retrieval, ranking, and generation, allowing learners to understand each stage independently and experiment with different retrieval strategies without modifying the generation component

vs others: More transparent than using LangChain RAG chains because it shows the underlying retrieval and ranking logic explicitly, enabling customization and debugging of retrieval quality rather than treating it as a black box

via “contextual knowledge retrieval”

Qwen3.6-Plus: Towards real world agents

Unique: Combines RAG with a context-aware indexing system, ensuring that responses are not only accurate but also contextually relevant.

vs others: More accurate than standard search engines, as it tailors results based on user context and intent.

via “retrieval-augmented-generation-system-resource-mapping”

A curated list of Generative AI tools, works, models, and references

Unique: Treats RAG as a distinct capability with dedicated resources covering the full pipeline (embeddings → vector databases → retrieval → reranking), rather than treating it as an LLM application pattern. Recognizes that RAG requires specialized infrastructure (vector databases, embedding models) beyond base LLMs

vs others: More comprehensive than single-tool documentation (Pinecone, Weaviate) by covering the full RAG ecosystem, but less detailed than specialized communities (Hugging Face, Papers with Code) which provide benchmarks and comparative analysis of retrieval methods

via “retrieval augmented generation (rag) technique documentation with architecture patterns”

🐙 Guides, papers, lessons, notebooks and resources for prompt engineering, context engineering, RAG, and AI Agents.

Unique: Positions RAG within the broader prompt engineering landscape, showing how it complements other techniques (CoT, few-shot prompting) and contrasts with alternatives (fine-tuning, in-context learning) rather than treating RAG in isolation

vs others: More comprehensive than vendor-specific RAG tutorials because it covers architectural principles independent of particular vector databases; more practical than academic RAG papers because it includes implementation patterns and integration strategies

via “self-correcting-generation-with-retrieval-feedback”

Agentic RAG is a different beast entirely.

Unique: Closes the loop between generation and retrieval by using agent reasoning to validate answers and trigger corrective actions, rather than treating generation as a one-shot process that assumes retrieved context is sufficient

vs others: More reliable than standard RAG because it actively detects and corrects hallucinations through validation feedback, whereas naive RAG generates once and trusts the LLM to stay grounded regardless of context quality

via “context-aware prompt augmentation with retrieved memories”

Hello HN! I built collabmem, a simple memory system for long-term collaboration between humans and AI assistants. And it's easy to install, just ask Claude Code: Install the long-term collaboration memory system by cloning https://github.com/visionscaper/collabmem to a te

Unique: Implements RAG specifically for collaborative memory, automatically surfacing relevant past interactions to inform current LLM responses without explicit user prompting, with token-aware memory selection

vs others: Automatically augments prompts with relevant memories unlike manual context injection, and uses semantic relevance ranking rather than keyword matching for memory selection

via “contextual retrieval for enhanced response generation”

Build and deploy pragmatic retrieval-augmented generation (RAG) agents efficiently. Integrate various data sources and APIs to enhance your AI agents' capabilities. Streamline agent development with a robust core library designed for practical applications.

Unique: Combines semantic and keyword-based retrieval methods to enhance the relevance of information accessed by RAG agents.

vs others: Delivers more contextually relevant outputs than standard RAG implementations that rely solely on keyword matching.

via “retrieval-augmented generation (rag) pipeline composition”

Community contributed LangChain integrations.

Unique: Provides pre-built RetrievalQA chains that combine document retrieval with LLM generation, supporting multiple retrieval strategies (similarity, MMR, ensemble). Chains handle source attribution and can be customized via composition.

vs others: More comprehensive than manual RAG implementation because it handles end-to-end pipelines, and more flexible than single-purpose RAG tools because it supports customization via chain composition.

via “memory-augmented inference with context retrieval and generation”

This package contains the code for training a memory-augmented GPT model on patient data. Please note that this is not the 'letta' company project with thehttps://github.com/letta-ai/letta; for use of their package, plsuse 'pymemgpt' instead.

Unique: Implements memory retrieval as a first-class inference component integrated into the model architecture rather than as post-processing; uses learned attention mechanisms to weight retrieved memory, allowing the model to learn context relevance during training

vs others: More efficient than naive RAG by integrating retrieval into model forward pass; learned memory weighting is more sophisticated than fixed retrieval strategies

via “semantic search and retrieval augmentation integration”

Opus 4.7 is the next generation of Anthropic's Opus family, built for long-running, asynchronous agents. Building on the coding and agentic strengths of Opus 4.6, it delivers stronger performance on...

Unique: Opus 4.7's 200K context window enables RAG patterns without complex chunking or hierarchical retrieval; model can reason over 50+ retrieved documents simultaneously, enabling more comprehensive synthesis than competitors limited to 10-20 documents

vs others: Enables RAG with longer context than GPT-4, reducing need for multi-stage retrieval pipelines; better at synthesizing insights across many documents due to extended context; integrates seamlessly with OpenRouter's retrieval partners

via “semantic search and retrieval augmentation”

GPT-5.4 is OpenAI’s latest frontier model, unifying the Codex and GPT lines into a single system. It features a 1M+ token context window (922K input, 128K output) with support for...

Unique: Native integration with major vector databases (Pinecone, Weaviate, Milvus) through standardized APIs eliminates custom adapter code; uses unified embedding space across retrieval and generation, ensuring semantic consistency between retrieved context and model responses

vs others: Faster than LangChain RAG pipelines (native integration vs. abstraction layer) and more flexible than Anthropic's context window approach (dynamic retrieval vs. static context); outperforms Gemini's retrieval augmentation on citation accuracy due to explicit document tracking