# Vectors & LLMs

A computing topic covering vector data and large language models, including retrieval-augmented generation and database indexing.

This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles.

## From Industrial Product Documentation to Automated Digital Product Passport Generation: An Experimental Study of Large Language Models

DevFeed: [From Industrial Product Documentation to Automated Digital Product Passport Generation: An Experimental Study of Large Language Models](<https://devfeed.tech/articles/from-industrial-product-documentation-to-automated-digital-product-passport-generation-an-experimental-study-of-large-language-models-145733.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.24998>)

Author: Marco Buecheler, Grant Richards, Nathan Hartman

Published: 2026-11-20T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Databases](<https://devfeed.tech/sources/arxiv-databases.md>), [arXiv: Software Engineering](<https://devfeed.tech/sources/arxiv-software-engineering.md>)

Topics: [Retrieval Augmented Generation (RAG)](<https://devfeed.tech/topics/retrieval-augmented-generation-rag.md>), [Vectors & LLMs](<https://devfeed.tech/topics/vectors-llms.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>), [hybrid-search](<https://devfeed.tech/topics/hybrid-search.md>), [Factuality scoring in AI](<https://devfeed.tech/topics/factuality-scoring-in-ai.md>)

Tags: [automated](<https://devfeed.tech/tags/automated.md>), [cs-db](<https://devfeed.tech/tags/cs-db.md>), [cs-se](<https://devfeed.tech/tags/cs-se.md>), [digital](<https://devfeed.tech/tags/digital.md>), [documentation](<https://devfeed.tech/tags/documentation.md>), [experimental](<https://devfeed.tech/tags/experimental.md>), [generation](<https://devfeed.tech/tags/generation.md>), [industrial](<https://devfeed.tech/tags/industrial.md>), [language-models](<https://devfeed.tech/tags/language-models.md>), [large-language-models](<https://devfeed.tech/tags/large-language-models.md>), [passport](<https://devfeed.tech/tags/passport.md>), [product](<https://devfeed.tech/tags/product.md>), [study](<https://devfeed.tech/tags/study.md>)

### AI overview

The study tests a RAG pipeline for generating Digital Product Passports as Asset Administration Shell instances from scattered industrial documentation. It chunks documents, stores embeddings in ChromaDB, combines semantic and lexical retrieval against AAS submodel requirements, and supplies retrieved content to five LLMs. Across 4,560 executions covering 76 product-submodel combinations, 97.4% of outputs were valid JSON, but 44.9% passed full AAS validation. Deterministic scoring found mean completeness of 49.6%, conditional accuracy of 58.3%, and overall field correctness of 30.8%; an LLM judge rated the same outputs higher at 66.5%, 88.9%, and 60.4%. The abstract says four prompt-guidance levels were evaluated but does not report their comparative effects.

### Source excerpt

arXiv:2611.24998v1 Announce Type: cross Abstract: Manufacturing companies that are planning to sell certain products in the European Union (EU) will be required to provide mandatory digital product passports (DPPs). The use of DPPs is to provide standardized product information throughout multiple product life cycles in a circular economy. Required passport information, such as technical specifications, material composition, and environmental product footprint, is available in manufacturers' data sources but is scattered across different heterogeneous systems, documents, and data formats. The given paper investigates using large language models (LLMs) to automate the generation of DPPs based on the asset administration shell (AAS). A retrieval-augmented generation (RAG) approach is used to extract product information from industrial documentation and generate populated AAS instances. Product documentation is processed and divided into text chunks. The text chunks are then stored in a persistent ChromaDB vector database using vector embeddings. The retrieval process combines semantic and lexical search to identify relevant information based on standardized AAS submodel template requirements. The retrieved information is provided to five LLMs to generate structured AAS instances. The experiments cover different industrial products and AAS submodels. Four levels of prompt guidance are investigated to assess their influence on generation quality. The generated passports are evaluated against a ground truth. The experimental evaluation includes 4,560 executions across 76 product-submodel combinations. Of the generated passport outputs, 97.4% are valid JSON, while 44.9% pass the full AAS validation procedure. Under deterministic scoring, mean completeness, conditional accuracy, and overall field correctness are 49.6%, 58.3%, and 30.8%, respectively. An LLM-as-a-judge yields better scores of 66.5%, 88.9%, and 60.4% for the same outputs.

## A Neurosymbolic Multimodal Framework for 30-Day Diabetic Readmission Prediction: Integrating Structured EHR Data, Dense Clinical Embeddings, and Knowledge-Grounded Symbolic Extraction

DevFeed: [A Neurosymbolic Multimodal Framework for 30-Day Diabetic Readmission Prediction: Integrating Structured EHR Data, Dense Clinical Embeddings, and Knowledge-Grounded Symbolic Extraction](<https://devfeed.tech/articles/a-neurosymbolic-multimodal-framework-for-30-day-diabetic-readmission-prediction-integrating-structured-ehr-data-dense-clinical-embeddings-and-knowledge-grounded-symbolic-145488.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.25236>)

Author: Una Suman Kumar Patro, Sibangi Subhadarsani, Narendra Badam

Published: 2026-11-20T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [Vectors & LLMs](<https://devfeed.tech/topics/vectors-llms.md>), [Embeddings](<https://devfeed.tech/topics/embeddings.md>), [Representation learning](<https://devfeed.tech/topics/representation-learning.md>), [AI Integration Strategies](<https://devfeed.tech/topics/ai-integration-strategies.md>), [AI Chat](<https://devfeed.tech/topics/ai-chat.md>), [AI for data automation](<https://devfeed.tech/topics/ai-for-data-automation.md>), [data-engineering](<https://devfeed.tech/topics/data-engineering.md>)

Tags: [cross-encoder](<https://devfeed.tech/tags/cross-encoder.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [dataset](<https://devfeed.tech/tags/dataset.md>), [embeddings](<https://devfeed.tech/tags/embeddings.md>), [hallucinations](<https://devfeed.tech/tags/hallucinations.md>), [multimodal](<https://devfeed.tech/tags/multimodal.md>), [retrieval-augmented-generation](<https://devfeed.tech/tags/retrieval-augmented-generation.md>), [structured](<https://devfeed.tech/tags/structured.md>), [xgboost](<https://devfeed.tech/tags/xgboost.md>)

### AI overview

On a held-out cohort of 20,153 encounters, the multimodal model improved sensitivity over the structured EHR baseline (0.6667 vs. 0.5839), achieved the best F2-score (0.3824) and lower Brier score (0.0977 vs. 0.1080), but had slightly lower AUROC (0.6375 vs. 0.6477) and specificity (0.5161). It combines engineered structured features, 32 compressed Bio_ClinicalBERT embedding components, and symbolic entities extracted through hybrid retrieval and negation auditing, then predicts with cost-sensitive XGBoost. In a separate 500-note code-switched benchmark, grounded retrieval raised Hinglish symptom recall and precision and reduced entity hallucinations. Prior utilization was the dominant risk signal. The reported findings are from the abstract's dataset and evaluations.

### Source excerpt

arXiv:2611.25236v1 Announce Type: new Abstract: Unplanned 30-day readmissions among patients with diabetes mellitus impose a substantial clinical and economic burden, yet most predictive models use only structured electronic health record (EHR) data and ignore free-text clinical notes, which often contain code-mixed Hindi-English (Hinglish) phrasing and negation that general-purpose language models parse unreliably. We propose a neurosymbolic multimodal framework evaluated on the public Diabetes 130-US Hospitals dataset (101,766 encounters; 71,518 patients) with a patient-level grouped 80/20 split. Three streams are fused: engineered structured features; 32 components of Bio_ClinicalBERT note embeddings obtained by truncated singular value decomposition; and symbolic entities extracted by an open-weight clinical language model within a three-stage hybrid retrieval-augmented generation pipeline (dense retrieval, sparse lexical retrieval, and cross-encoder re-ranking) with reflective negation auditing. A cost-sensitive XGBoost classifier with a decision threshold calibrated by Youden's J statistic produces the prediction. On the held-out test cohort (n = 20,153), the structured baseline achieved an area under the receiver operating characteristic curve (AUROC) of 0.6477, sensitivity of 0.5839, and Brier score of 0.1080. The full multimodal model achieved the highest sensitivity (0.6667; +8.28 percentage points), the best F2-score (0.3824), and the lowest Brier score (0.0977), at a slightly lower AUROC (0.6375) and specificity (0.5161). Shapley additive explanations showed that prior utilization dominates risk, with compressed narrative components among the top predictors. On a 500-note code-switched benchmark, grounded retrieval raised Hinglish symptom recall from 4.8% to 40.2% and precision from 3.1% to 45.2%, while reducing entity hallucinations from 1.74 to 0.00 per note.

## What Is a Vector Database? How It Works and Why AI Needs One

DevFeed: [What Is a Vector Database? How It Works and Why AI Needs One](<https://devfeed.tech/articles/what-is-a-vector-database-how-it-works-and-why-ai-needs-one-145883.md>)

Original publisher: [Read original article](<https://montecarlo.ai/blog-what-is-a-vector-database>)

Author: Virna Sekuj

Published: 2026-11-19T17:18:16Z

Content type: article

Language: en

Sources: [Monte Carlo](<https://devfeed.tech/sources/monte-carlo.md>)

Topics: [Vectors & LLMs](<https://devfeed.tech/topics/vectors-llms.md>), [Embeddings](<https://devfeed.tech/topics/embeddings.md>), [vector embeddings in AI](<https://devfeed.tech/topics/vector-embeddings-in-ai.md>), [Retrieval Augmented Generation (RAG)](<https://devfeed.tech/topics/retrieval-augmented-generation-rag.md>), [ann](<https://devfeed.tech/topics/ann.md>), [Database](<https://devfeed.tech/topics/database.md>), [information retrieval](<https://devfeed.tech/topics/information-retrieval.md>), [hybrid-search](<https://devfeed.tech/topics/hybrid-search.md>)

Tags: [agent-trust](<https://devfeed.tech/tags/agent-trust.md>), [agent-trust-glossary](<https://devfeed.tech/tags/agent-trust-glossary.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [data](<https://devfeed.tech/tags/data.md>), [database](<https://devfeed.tech/tags/database.md>), [embeddings](<https://devfeed.tech/tags/embeddings.md>), [glossary](<https://devfeed.tech/tags/glossary.md>), [rag](<https://devfeed.tech/tags/rag.md>), [rag-systems](<https://devfeed.tech/tags/rag-systems.md>), [retrieval](<https://devfeed.tech/tags/retrieval.md>), [retrieval-augmented-generation](<https://devfeed.tech/tags/retrieval-augmented-generation.md>), [retrieval-augmented-generation-rag](<https://devfeed.tech/tags/retrieval-augmented-generation-rag.md>), [vector](<https://devfeed.tech/tags/vector.md>), [vector-database](<https://devfeed.tech/tags/vector-database.md>), [vector-database-examples](<https://devfeed.tech/tags/vector-database-examples.md>), [vector-database-for-rag](<https://devfeed.tech/tags/vector-database-for-rag.md>), [vector-database-vs-relational-database](<https://devfeed.tech/tags/vector-database-vs-relational-database.md>), [vector-search](<https://devfeed.tech/tags/vector-search.md>), [vector-store](<https://devfeed.tech/tags/vector-store.md>)

### AI overview

Vector databases store content embeddings and use similarity search to retrieve relevant passages even when query wording differs. In a typical RAG flow, source content is chunked and embedded, stored with metadata, and indexed for approximate nearest-neighbor search; the user query is embedded and matching chunks are passed to the LLM as context. ANN indexing trades a small amount of precision for speed, and metadata filters can narrow results. Options include dedicated databases, database extensions, and vector search built into data platforms; choice depends on scale, latency, data location, and operational overhead. Retrieval can silently degrade when source syncs become stale, ingestion or chunking breaks, embedding models change without re-embedding, duplicate versions compete, or permission filters are missing. The text recommends monitoring freshness and ingestion volume, checking chunk quality, tracking embedding versions, measuring relevance and grounding, and tracing which chunks each agent request retrieves.

### Source excerpt

A vector database is a database built to store embeddings, which are numeric representations of the meaning of text, images, or other data, and to quickly find the embeddings most similar to a query. It's the retrieval layer behind most retrieval-augmented generation (RAG) systems and many AI agents, letting them find relevant information by meaning instead of exact keywords.

## Making Mathematical Knowledge Explainable, Accessible and Interoperable Through Large Language Model Integration

DevFeed: [Making Mathematical Knowledge Explainable, Accessible and Interoperable Through Large Language Model Integration](<https://devfeed.tech/articles/making-mathematical-knowledge-explainable-accessible-and-interoperable-through-large-language-model-integration-142545.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2607.24512>)

Author: Jan Range, Bj\\"orn Schembera, Dominik G\\"oddeke

Published: 2026-11-19T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [Large Language Model](<https://devfeed.tech/topics/llm.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [AI Integration Strategies](<https://devfeed.tech/topics/ai-integration-strategies.md>), [Vectors & LLMs](<https://devfeed.tech/topics/vectors-llms.md>), [knowledge-engineering](<https://devfeed.tech/topics/knowledge-engineering.md>), [Retrieval Augmented Generation (RAG)](<https://devfeed.tech/topics/retrieval-augmented-generation-rag.md>), [Wikidata](<https://devfeed.tech/topics/wikidata.md>)

Tags: [accessible](<https://devfeed.tech/tags/accessible.md>), [benefits](<https://devfeed.tech/tags/benefits.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-dl](<https://devfeed.tech/tags/cs-dl.md>), [explainability](<https://devfeed.tech/tags/explainability.md>), [infrastructure](<https://devfeed.tech/tags/infrastructure.md>), [interaction](<https://devfeed.tech/tags/interaction.md>), [interoperability](<https://devfeed.tech/tags/interoperability.md>), [join](<https://devfeed.tech/tags/join.md>), [knowledge-base](<https://devfeed.tech/tags/knowledge-base.md>), [knowledge-graph](<https://devfeed.tech/tags/knowledge-graph.md>), [large-language-model](<https://devfeed.tech/tags/large-language-model.md>), [mcp](<https://devfeed.tech/tags/mcp.md>), [mcp-protocol](<https://devfeed.tech/tags/mcp-protocol.md>), [mcp-server](<https://devfeed.tech/tags/mcp-server.md>), [metadata](<https://devfeed.tech/tags/metadata.md>), [model](<https://devfeed.tech/tags/model.md>), [model-context-protocol](<https://devfeed.tech/tags/model-context-protocol.md>), [model-context-protocol-mcp](<https://devfeed.tech/tags/model-context-protocol-mcp.md>), [models](<https://devfeed.tech/tags/models.md>), [natural-language](<https://devfeed.tech/tags/natural-language.md>), [retrieval](<https://devfeed.tech/tags/retrieval.md>)

### AI overview

MathModDB, a curated mathematical-model knowledge base built on Wikibase, can be difficult to use without navigating its complex interface or knowing SPARQL and Wikibase APIs; linking its records with research data in repositories such as Dataverse is also challenging. The paper proposes an MCP server that connects LLM dialogue to MathModDB using vector-indexed schema retrieval and a Steiner-tree-based join planner. The authors report that this approach improves accessibility and explainability while grounding LLM use in curated knowledge, and can simplify interoperability with external databases and tools. They present two use cases and say the architecture can extend to other Wikibase-based systems. The supplied text is an abstract, so it does not detail the use cases or evaluate performance.

### Source excerpt

arXiv:2607.24512v2 Announce Type: replace Abstract: Mathematical models are central to formalizing research problems, yet their documentation often falls short of FAIR principles. Knowledge bases such as the Mathematical Model Database (MathModDB) address this gap by providing curated, semantically rich representations of mathematical models. Built on Wikibase, the same open-source infrastructure underlying Wikidata, MathModDB utilizes Semantic Web technologies to support Linked Open Data, collaborative editing, and the storage of semantically enriched metadata, making it a domain-specific knowledge graph within the broader Wikidata ecosystem. However, access to MathModDB currently requires either navigating a complex web interface or proficiency in SPARQL and Wikibase APIs, posing significant barriers for potential users. In addition, the combination of such curated knowledge bases with actual research data stored, e.g., in Dataverse repository instances, remains a challenge. To overcome these limitations, we propose integrating Large Language Models (LLMs) with MathModDB via a Model Context Protocol (MCP) server that exposes a vector-indexed schema retrieval and Steiner-tree-based join planner, combining dialogue-based natural language interaction with curated, epistemically grounded knowledge. Although instantiated on MathModDB, the architecture can be applied to other Wikibase-based systems. We demonstrate that this approach enables epistemically grounded LLM usage, improves model explainability and accessibility beyond what the standard Wikibase interface offers, and simplifies interoperability with external databases and tools, such as Dataverse data repositories. We illustrate the benefits of combining the accessibility of an LLM with the epistemic safety of a curated knowledge base through the adaptability of the MCP protocol by two use cases.

## SIMQ: A Single Graph-based Index for Multi-Vector Query

DevFeed: [SIMQ: A Single Graph-based Index for Multi-Vector Query](<https://devfeed.tech/articles/simq-a-single-graph-based-index-for-multi-vector-query-140627.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.22079>)

Author: Yifan Lu, Jiabao Jin, Xiaoyao Zhong, Peng Cheng, Xiangyu Wang, Haotian Li, Hua Chen, Junqing Chen, Wei Xi, Heng Tao Shen

Published: 2026-11-18T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Databases](<https://devfeed.tech/sources/arxiv-databases.md>)

Topics: [ann](<https://devfeed.tech/topics/ann.md>), [clustering](<https://devfeed.tech/topics/clustering.md>), [Vectors & LLMs](<https://devfeed.tech/topics/vectors-llms.md>)

Tags: [abstract](<https://devfeed.tech/tags/abstract.md>), [accelerated](<https://devfeed.tech/tags/accelerated.md>), [analysis](<https://devfeed.tech/tags/analysis.md>), [approximation](<https://devfeed.tech/tags/approximation.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [benchmarks](<https://devfeed.tech/tags/benchmarks.md>), [bottlenecks](<https://devfeed.tech/tags/bottlenecks.md>), [candidate](<https://devfeed.tech/tags/candidate.md>), [cluster](<https://devfeed.tech/tags/cluster.md>), [clustering](<https://devfeed.tech/tags/clustering.md>), [clusters](<https://devfeed.tech/tags/clusters.md>), [compression](<https://devfeed.tech/tags/compression.md>), [construction](<https://devfeed.tech/tags/construction.md>), [cost](<https://devfeed.tech/tags/cost.md>), [cpu](<https://devfeed.tech/tags/cpu.md>), [cs-db](<https://devfeed.tech/tags/cs-db.md>), [cs-ir](<https://devfeed.tech/tags/cs-ir.md>), [data](<https://devfeed.tech/tags/data.md>), [graph](<https://devfeed.tech/tags/graph.md>), [index](<https://devfeed.tech/tags/index.md>), [query](<https://devfeed.tech/tags/query.md>), [vector](<https://devfeed.tech/tags/vector.md>)

### AI overview

SIMQ approximates ColBERT-style MaxSim retrieval by grouping token vectors into fine-grained, data-aware clusters, indexing their centroids in one graph, and navigating that graph per query token. The approach avoids building a full centroid interaction matrix and does not require GPU acceleration for this search process; the index is independent of PLAID's residual representation. The paper's theoretical result is conditional on a manifold assumption and successful cell retrieval, under which data-aware partitioning has lower expected approximation error than data-oblivious partitioning. SIMQ+ adds count-based scoring and a dynamic candidate set to improve robustness across query conditions. Evaluation on five benchmarks compares recall, throughput, and construction time against several baselines, primarily on CPU; large-scale PLAID k-means centroids were built on a separate GPU machine. SIMQ is integrated into VSAG.

### Source excerpt

arXiv:2611.22079v1 Announce Type: cross Abstract: Multi-vector retrieval models such as ColBERT achieve state-of-the-art effectiveness by representing documents as sets of token-level vectors and scoring via MaxSim, but their high computational cost limits scalability to large corpora. Existing approximate methods face fundamental bottlenecks: centroid interaction matrix approaches require computing a full matrix at every query and rely on GPU-accelerated clustering, while data-oblivious compression methods yield unstable recall that requires repeated compensation. We propose SIMQ (Single graph Index for Multi-vector Query), a framework that bridges single-vector graph indices and multi-vector retrieval through data-aware sampling-based clustering and per-token graph navigation, simultaneously bypassing all aforementioned bottlenecks. SIMQ organizes all token vectors into fine-grained clusters via data-aware sampling, builds a single graph index over cluster centroids independently of PLAID's residual representation, and approximates the MaxSim score through per-token graph search without materializing any interaction matrix or relying on GPU acceleration. We provide theoretical analysis showing that, under a manifold assumption and successful cell retrieval, data-aware partitioning has a smaller expected approximation error than a data-oblivious partition. Building on this foundation, SIMQ+ further introduces count-based scoring and a dynamic candidate set to improve robustness across varying query conditions. Experiments on five benchmarks compare recall, throughput, and construction time against several baselines, primarily on CPU; large-scale PLAID k-means centroids are built on a separate GPU machine. SIMQ has been integrated into VSAG, an Ant Group's open-source vector search library.

## The Illusion of Grounding: Do Vision-Language Models Weigh Retrieved Text and Images Like an Ideal Observer?

DevFeed: [The Illusion of Grounding: Do Vision-Language Models Weigh Retrieved Text and Images Like an Ideal Observer?](<https://devfeed.tech/articles/the-illusion-of-grounding-do-vision-language-models-weigh-retrieved-text-and-images-like-an-ideal-observer-141848.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.23127>)

Author: Basavaraj R M, Harshal Chandel

Published: 2026-11-18T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [vlm](<https://devfeed.tech/topics/vlm.md>), [multimodal-ai](<https://devfeed.tech/topics/multimodal-ai.md>), [Vectors & LLMs](<https://devfeed.tech/topics/vectors-llms.md>)

Tags: [audit](<https://devfeed.tech/tags/audit.md>), [bayesian](<https://devfeed.tech/tags/bayesian.md>), [bias](<https://devfeed.tech/tags/bias.md>), [capture](<https://devfeed.tech/tags/capture.md>), [coco](<https://devfeed.tech/tags/coco.md>), [code](<https://devfeed.tech/tags/code.md>), [consistency](<https://devfeed.tech/tags/consistency.md>), [coverage](<https://devfeed.tech/tags/coverage.md>), [cs-cl](<https://devfeed.tech/tags/cs-cl.md>), [cs-cv](<https://devfeed.tech/tags/cs-cv.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [cue](<https://devfeed.tech/tags/cue.md>), [error](<https://devfeed.tech/tags/error.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [gaussian](<https://devfeed.tech/tags/gaussian.md>), [grounding](<https://devfeed.tech/tags/grounding.md>), [images](<https://devfeed.tech/tags/images.md>), [language](<https://devfeed.tech/tags/language.md>), [language-models](<https://devfeed.tech/tags/language-models.md>), [models](<https://devfeed.tech/tags/models.md>), [text](<https://devfeed.tech/tags/text.md>), [vision](<https://devfeed.tech/tags/vision.md>), [vision-language-models](<https://devfeed.tech/tags/vision-language-models.md>)

### AI overview

On clean inputs, four evaluated open vision-language models achieve 92.5%-94.2% accuracy, but under retrieval or image stress accuracy drops to 41.3%-59.9%. The study introduces GIB-200K, a 200,000-item benchmark built from COCO and Visual Genome with controlled retrieval noise, visual degradation, and cross-modal conflicts, and methods to estimate modality weighting and measure answers not grounded in images. The two LLaVA models overweight retrieved text, while InternVL3 favors image evidence; 34.8%-71.4% of clean-correct answers are abandoned when documents are misleading. A training-free check comparing answers with and without retrieval trades coverage for accuracy under misleading documents, while image-only fallback fails when images are heavily degraded. Results cover one seeded 2,000-item split.

### Source excerpt

arXiv:2611.23127v1 Announce Type: new Abstract: Retrieval-augmented vision-language models (VLMs) answer from two sources of evidence that can each fail: the retrieved text can be wrong or noisy, and the image can be degraded or contradicted. We introduce GIB-200K, 200,000 image-question items built from COCO and Visual Genome, each expanded into controlled stress conditions: hard-negative retrieval noise (1-20 documents), visual ambiguity (Gaussian and motion blur, low-resource crops) and cross-modal conflict with counterfactual flip pairs. Instead of only measuring degradation, we ask how models combine the two modalities. From ideal Bayesian cue combination we derive a testable prediction that yields formal definitions of Knowledge Inertia (text over-weighted relative to the model's own image evidence) and Sensory Capture (the reverse), and we estimate the weights with a measurement-error-corrected regression whose recovery of known weights we verify. A counterfactual flip decomposition separates modality bias from concept priors, and a Grounding Illusion Score measures how much clean accuracy is not grounded in the image. We evaluate 4 open models on one frozen, seeded evaluation split of 2,000 GIB-200K items. All models are accurate on clean inputs (92.5%-94.2%) but adverse accuracy falls to 41.3%-59.9%. The two LLaVA models over-weight the text (Knowledge Inertia, w_T/w_I = 1.229 and 1.322), whereas InternVL3 favours the image (w_T/w_I = 0.680); between 34.8% and 71.4% of the clean answers that are correct are abandoned when the document lies. A training-free consistency check that compares the answer with and without retrieval trades coverage for accuracy under misleading documents, while falling back to the image fails when the image is heavily degraded. All per-instance outputs, code, and an audit that recomputes every reported open-model number are released.

## Retrieval-Augmented Correction for Generalist Medical Image Segmentation

DevFeed: [Retrieval-Augmented Correction for Generalist Medical Image Segmentation](<https://devfeed.tech/articles/retrieval-augmented-correction-for-generalist-medical-image-segmentation-140128.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.23189>)

Author: Vincenzo Marcian\\\`o, Xiaoming Zhang, Michela Antonelli, Sebastien Ourselin, Maria A Zuluaga

Published: 2026-11-18T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [Medical imaging](<https://devfeed.tech/topics/medical-imaging.md>), [Enhancing AI model performance](<https://devfeed.tech/topics/enhancing-ai-model-performance.md>), [Vectors & LLMs](<https://devfeed.tech/topics/vectors-llms.md>), [Image](<https://devfeed.tech/topics/image.md>)

Tags: [context](<https://devfeed.tech/tags/context.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cv](<https://devfeed.tech/tags/cs-cv.md>), [dataset](<https://devfeed.tech/tags/dataset.md>), [embedding](<https://devfeed.tech/tags/embedding.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [generalist](<https://devfeed.tech/tags/generalist.md>), [input](<https://devfeed.tech/tags/input.md>), [knowledge-base](<https://devfeed.tech/tags/knowledge-base.md>), [mask](<https://devfeed.tech/tags/mask.md>), [medical](<https://devfeed.tech/tags/medical.md>), [multimodal](<https://devfeed.tech/tags/multimodal.md>), [public](<https://devfeed.tech/tags/public.md>), [retrieval](<https://devfeed.tech/tags/retrieval.md>), [similarity](<https://devfeed.tech/tags/similarity.md>), [ultrasound](<https://devfeed.tech/tags/ultrasound.md>)

### AI overview

GRACE refines medical image segmentation masks by retrieving verified prior cases from a shared knowledge base, which can serve as a shape prior. Its refiner is trained once across multiple anatomies; in a leave-one-anatomy-out experiment, verified cases for an unseen solid organ were added to the index while the refiner remained frozen. Across 18 public CT, MRI, and ultrasound datasets and seven upstream segmentors, GRACE added up to 0.35 DSC for weak inputs (baseline 0.386), while changes were small on a strong segmentor (baseline 0.932). Transfer to modalities absent from both training and the index was partial and anatomy-dependent, and remains an open research direction.

### Source excerpt

arXiv:2611.23189v1 Announce Type: cross Abstract: Automated segmentation is routinely used as the first step in quantitative image-based clinical workflows, where predicted masks are subsequently reviewed and corrected by human operators. Automated correction of the residual structural errors is currently addressed one anatomy at a time, so the number of trained correctors scales with the number of segmentation targets. We present GRACE, a retrieval-augmented refiner that edits an input mask under context drawn from a shared knowledge base of verified prior cases. The refiner is trained once on a multi-organ corpus spanning compact organs, cavities, branching vessels, and tumour lesions. In the held-out dataset-transfer evaluation, three held-out datasets query a fixed knowledge base built from training data only. A separate leave-one-anatomy-out experiment tests parameter-free index extension: verified cases of an unseen solid organ enter the knowledge base, and the refiner stays frozen. A light shape-aware projection head on a frozen multimodal encoder maps embedding similarity to the shape agreement of the masks, so the knowledge base returns cases that the refiner can use as a shape prior. We evaluate GRACE on 18 public datasets spanning CT, MRI and ultrasound, using nine refinement configurations and seven upstream segmentors that span weak, mid, and strong operating regimes. We report each change with the DSC of its input mask. On the weak tier, the input DSC is 0.386 and GRACE adds up to +0.35 DSC. On a strong segmentor, the input DSC is 0.932 and the mean change is small. Transfer to an imaging modality that is absent from training and from the index is partial and depends on the anatomy, and remains an open research direction. Our results suggest retrieval-conditioned refinement is a practical alternative to anatomy-specific correction models. Code: https://github.com/robustml-eurecom/grace

## Add your organization's runbooks and policies using Retrieval Augmented Generation (RAG) on Amazon Aurora PostgreSQL

DevFeed: [Add your organization's runbooks and policies using Retrieval Augmented Generation (RAG) on Amazon Aurora PostgreSQL](<https://devfeed.tech/articles/add-your-organization-s-runbooks-and-policies-using-retrieval-augmented-generation-rag-on-amazon-aurora-postgresql-138650.md>)

Original publisher: [Read original article](<https://aws.amazon.com/blogs/database/add-your-organizations-runbooks-and-policies-using-retrieval-augmented-generation-rag-on-amazon-aurora-postgresql/>)

Author: Narendra Tawar

Published: 2026-11-17T19:43:40Z

Content type: tutorial

Language: en

Sources: [AWS Database Blog](<https://devfeed.tech/sources/aws-database-blog.md>)

Topics: [Amazon Bedrock Knowledge Bases](<https://devfeed.tech/topics/amazon-bedrock-knowledge-bases.md>), [Retrieval Augmented Generation (RAG)](<https://devfeed.tech/topics/retrieval-augmented-generation-rag.md>), [Amazon Aurora](<https://devfeed.tech/topics/amazon-aurora.md>), [Vectors & LLMs](<https://devfeed.tech/topics/vectors-llms.md>), [Enhancing AI model performance](<https://devfeed.tech/topics/enhancing-ai-model-performance.md>), [Amazon S3](<https://devfeed.tech/topics/amazon-s3.md>), [Access Control](<https://devfeed.tech/topics/access-control.md>)

Tags: [advanced-300](<https://devfeed.tech/tags/advanced-300.md>), [amazon-aurora](<https://devfeed.tech/tags/amazon-aurora.md>), [amazon-bedrock](<https://devfeed.tech/tags/amazon-bedrock.md>), [amazon-bedrock-knowledge-bases](<https://devfeed.tech/tags/amazon-bedrock-knowledge-bases.md>), [amazon-s3](<https://devfeed.tech/tags/amazon-s3.md>), [kiro](<https://devfeed.tech/tags/kiro.md>), [pgvector](<https://devfeed.tech/tags/pgvector.md>), [policies](<https://devfeed.tech/tags/policies.md>), [postgresql](<https://devfeed.tech/tags/postgresql.md>), [rag](<https://devfeed.tech/tags/rag.md>), [retrieval](<https://devfeed.tech/tags/retrieval.md>), [retrieval-augmented-generation-rag](<https://devfeed.tech/tags/retrieval-augmented-generation-rag.md>), [technical-how-to](<https://devfeed.tech/tags/technical-how-to.md>)

### AI overview

The walkthrough connects a Kiro DBA assistant to team runbooks and policies using Amazon Bedrock Knowledge Bases, Amazon S3, and Aurora PostgreSQL with pgvector. Bedrock chunks documents, creates Titan Text Embeddings V2 vectors, and stores them with text in Aurora; at query time, the Retrieval MCP server returns relevant chunks for the assistant to ground its recommendations. Setup requires Aurora PostgreSQL 15.4 or later, pgvector, the RDS Data API, and appropriate Bedrock, S3, and Secrets Manager access. The Knowledge Base needs the `mcp-multirag-kb=true` tag for automatic MCP discovery. For structured policies, section-based custom chunking may improve retrieval precision over default chunking. In the example, retrieval surfaces required peer review, CAB approval, and a manual snapshot before schema changes, while suggesting reader-endpoint traffic redirection as an immediate mitigation. The article also stresses that steering files are soft guardrails; read-only database permissions and IAM deny policies provide additional write protection. Re-sync documents as they change, and restrict `bedrock:Retrieve` access to protect sensitive runbooks.

### Source excerpt

In Part 2 of this series, you connect your Kiro DBA assistant to your team's runbooks and policies. Store runbooks in Amazon S3, index them with Amazon Bedrock Knowledge Bases on Amazon Aurora PostgreSQL using pgvector, and wire up retrieval so the assistant grounds its recommendations in your own documentation.

## Simplify AI retrieval with embedding generation using Microsoft-Embedding-1

DevFeed: [Simplify AI retrieval with embedding generation using Microsoft-Embedding-1](<https://devfeed.tech/articles/simplify-ai-retrieval-with-embedding-generation-using-microsoft-embedding-1-139451.md>)

Original publisher: [Read original article](<https://devblogs.microsoft.com/cosmosdb/microsoft-embedding-one/>)

Author: Khelan Modi

Published: 2026-11-17T18:45:11Z

Content type: release

Language: en

Sources: [Azure Cosmos DB Blog](<https://devfeed.tech/sources/azure-cosmos-db-blog.md>)

Topics: [information retrieval](<https://devfeed.tech/topics/information-retrieval.md>), [Vectors & LLMs](<https://devfeed.tech/topics/vectors-llms.md>), [AI-driven information retrieval](<https://devfeed.tech/topics/ai-driven-information-retrieval.md>), [Azure](<https://devfeed.tech/topics/azure.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [analyst](<https://devfeed.tech/tags/analyst.md>), [azure](<https://devfeed.tech/tags/azure.md>), [azure-cosmos-db](<https://devfeed.tech/tags/azure-cosmos-db.md>), [charts](<https://devfeed.tech/tags/charts.md>), [cosmos](<https://devfeed.tech/tags/cosmos.md>), [embedding](<https://devfeed.tech/tags/embedding.md>), [enterprise](<https://devfeed.tech/tags/enterprise.md>), [financial](<https://devfeed.tech/tags/financial.md>), [generation](<https://devfeed.tech/tags/generation.md>), [knowledge](<https://devfeed.tech/tags/knowledge.md>), [microsoft](<https://devfeed.tech/tags/microsoft.md>), [pages](<https://devfeed.tech/tags/pages.md>), [q3](<https://devfeed.tech/tags/q3.md>), [region](<https://devfeed.tech/tags/region.md>), [reports](<https://devfeed.tech/tags/reports.md>), [retrieval](<https://devfeed.tech/tags/retrieval.md>), [revenue](<https://devfeed.tech/tags/revenue.md>), [search](<https://devfeed.tech/tags/search.md>), [tables](<https://devfeed.tech/tags/tables.md>)

### AI overview

Microsoft Embedding 1 creates embeddings for text, images, and document pages in a shared vector space, allowing text queries to retrieve answers in charts, diagrams, and scanned files. The source reports a #1 ranking in its ViDoRe v3 size class (66.24 nDCG@10, as of November 16, 2026). It offers dense and late-interaction multi-vector output: multi-vector retrieval improves visual-document relevance, but can use substantial storage and requires MaxSim scoring. Users can limit vector count and dimensions; 32 vectors of 128 dimensions reportedly gives up about 3.4 nDCG@10 points while reducing storage roughly 300-fold versus unconstrained output. Dense vectors are positioned for text-heavy content, multi-vector for visual documents. The article describes enabling inference on a Cosmos DB account, generating embeddings with preview Python or .NET clients, then storing and searching vectors; input tokens are billed. It says text retrieval scores on par with text-embedding-3-large on BEIR, with smaller vectors, while benchmark results and claims are source-reported.

### Source excerpt

Most enterprise knowledge doesn't live in clean text. It lives in charts, tables, scanned pages, and diagrams, where text-only search often misses it: Financial reports: An analyst asks, "Which region's revenue fell in Q3?" The answer is a bar in a chart on page 14, and no sentence on the page says it. Technical manuals: A [...] The post Simplify AI retrieval with embedding generation using Microsoft-Embedding-1 appeared first on Azure Cosmos DB Blog.

## OpenClassGen: A Large-Scale Corpus of Real-World Python Classes for LLM Research

DevFeed: [OpenClassGen: A Large-Scale Corpus of Real-World Python Classes for LLM Research](<https://devfeed.tech/articles/openclassgen-a-large-scale-corpus-of-real-world-python-classes-for-llm-research-137911.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2504.15564>)

Author: Musfiqur Rahman, SayedHassan Khatoonabadi, Emad Shihab

Published: 2026-11-17T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Software Engineering](<https://devfeed.tech/sources/arxiv-software-engineering.md>)

Topics: [managing large-scale text data in AI](<https://devfeed.tech/topics/managing-large-scale-text-data-in-ai.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>), [Vectors & LLMs](<https://devfeed.tech/topics/vectors-llms.md>), [qwen](<https://devfeed.tech/topics/qwen.md>)

Tags: [abstract](<https://devfeed.tech/tags/abstract.md>), [analysis](<https://devfeed.tech/tags/analysis.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [benchmarks](<https://devfeed.tech/tags/benchmarks.md>), [branch](<https://devfeed.tech/tags/branch.md>), [but](<https://devfeed.tech/tags/but.md>), [capabilities](<https://devfeed.tech/tags/capabilities.md>), [cases](<https://devfeed.tech/tags/cases.md>), [class](<https://devfeed.tech/tags/class.md>), [classes](<https://devfeed.tech/tags/classes.md>), [claude](<https://devfeed.tech/tags/claude.md>), [code](<https://devfeed.tech/tags/code.md>), [code-generation](<https://devfeed.tech/tags/code-generation.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [cs-se](<https://devfeed.tech/tags/cs-se.md>), [large](<https://devfeed.tech/tags/large.md>), [llm](<https://devfeed.tech/tags/llm.md>), [python](<https://devfeed.tech/tags/python.md>), [real-world](<https://devfeed.tech/tags/real-world.md>), [research](<https://devfeed.tech/tags/research.md>), [scale](<https://devfeed.tech/tags/scale.md>), [world](<https://devfeed.tech/tags/world.md>)

### AI overview

OpenClassGen provides 324,843 human-written Python classes from 2,970 open-source projects, each paired with a self-contained generation skeleton of signatures and docstrings plus 27 static metrics. This addresses the small scale of prior class-level datasets and avoids requiring repository context to resolve specifications. In a curated executable subset of 300 classes with tests reaching 58% branch coverage, GPT-o4-mini, Claude-4-Sonnet, and Qwen-3-Coder achieved a CodeBERTScore-F3 of 0.89 but a functional pass rate of 0.33, with substantial variation across models. The corpus is positioned for training and analysis; these evaluation results apply to the curated subset, not necessarily the full dataset.

### Source excerpt

arXiv:2504.15564v4 Announce Type: replace Abstract: Existing class-level code generation datasets are either synthetic (ClassEval: 100 classes) or insufficient in scale for modern training needs (RealClassEval: 400 classes), hindering robust evaluation and empirical analysis. We present OpenClassGen, a large-scale corpus of 324,843 Python classes extracted from 2,970 engineered open-source projects. Each entry pairs a human-written class with its corresponding skeleton, which comprises class and method signatures with associated docstrings, and is enriched with 27 static code metrics covering complexity, coupling, cohesion, and inheritance properties. Unlike prior benchmarks that require repository-level context resolution, OpenClassGen provides self-contained class skeletons that serve as complete generation specifications. We demonstrate the corpus's utility by evaluating three LLMs (GPT-o4-mini, Claude-4-Sonnet, Qwen-3-Coder) on a curated, executable subset of 300 classes, enriched with test suites achieving 58% branch coverage. Results show strong semantic similarity (CodeBERTScore-F3: 0.89) but moderate functional correctness (pass rate: 0.33), with substantial variance across models. This variance, along with diverse class characteristics, confirms that OpenClassGen enables meaningful differentiation of LLM capabilities. The dataset supports diverse use cases, including fine-tuning, retrieval-augmented generation, difficulty modelling, and failure mode analysis. The complete dataset and curation scripts are publicly available at https://zenodo.org/records/18409150.

## AdaMenu: Query-Adaptive Menus for Decision Models over Large Label Spaces

DevFeed: [AdaMenu: Query-Adaptive Menus for Decision Models over Large Label Spaces](<https://devfeed.tech/articles/adamenu-query-adaptive-menus-for-decision-models-over-large-label-spaces-138816.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.19005>)

Author: Jiangze Yan, Yi Shen, Jieyun Huang, Wenjing Zhang, Zhaoxiang Liu, Ning Wang, Kai Wang, Shiguo Lian

Published: 2026-11-17T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [enhancing AI model accuracy](<https://devfeed.tech/topics/enhancing-ai-model-accuracy.md>), [Vectors & LLMs](<https://devfeed.tech/topics/vectors-llms.md>), [decision-making](<https://devfeed.tech/topics/decision-making.md>)

Tags: [adaptive](<https://devfeed.tech/tags/adaptive.md>), [beam-search](<https://devfeed.tech/tags/beam-search.md>), [classification](<https://devfeed.tech/tags/classification.md>), [cs-cl](<https://devfeed.tech/tags/cs-cl.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [decision-models](<https://devfeed.tech/tags/decision-models.md>), [error-propagation](<https://devfeed.tech/tags/error-propagation.md>), [limits](<https://devfeed.tech/tags/limits.md>), [retrieval](<https://devfeed.tech/tags/retrieval.md>), [size](<https://devfeed.tech/tags/size.md>), [token-usage](<https://devfeed.tech/tags/token-usage.md>)

### AI overview

AdaMenu handles large classification label sets by retrieving labeled examples to build a query-adaptive candidate menu, then making one decision-model request. Its default setting also adjusts menu size using aggregated retrieval scores. Across four label sets with up to 1,016 labels, it beat TypeSafe's beam-search baseline on all 16 model and label-set pairs where both ran, by 16.6-34.3 percentage points in accuracy. Against the full menu on the three smaller sets, accuracy ranged from 0.5 points lower to 7.2 points higher, while input token use fell by factors of 3.2-22.9. AdaMenu also ran in settings where option limits ruled out the full menu and the hierarchical baseline was unavailable.

### Source excerpt

arXiv:2611.19005v1 Announce Type: new Abstract: Decision models such as Jev return probabilities over caller-defined options and are increasingly used for classification. Large label spaces challenge both accuracy and scalability: larger menus can impair decisions, increase input costs, and exceed option limits. Hierarchical classification reduces menu size but introduces serial calls and error propagation. We propose AdaMenu, which builds a compact candidate menu from retrieved labeled examples and makes a single decision-model request. Its candidate labels adapt to the query, and its default setting also adjusts menu size using aggregated retrieval scores. We evaluate AdaMenu with open and closed decision models on four label sets containing up to 1,016 labels. It outperforms TypeSafe's beam-search baseline on all 16 model and label-set pairs where both methods run, improving accuracy by 16.6-34.3 percentage points. Relative to the full menu on the three smaller label sets, accuracy differences range from -0.5 to +7.2 percentage points, while input token usage falls by factors of 3.2-22.9. AdaMenu also runs when option limits exclude the full menu, including settings where the hierarchical baseline is unavailable.

## jXBW: A Compressed Index Enabling Structure-Aware JSONL Retrieval for Structured RAG

DevFeed: [jXBW: A Compressed Index Enabling Structure-Aware JSONL Retrieval for Structured RAG](<https://devfeed.tech/articles/jxbw-a-compressed-index-enabling-structure-aware-jsonl-retrieval-for-structured-rag-138277.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2508.12536>)

Author: Yasuo Tabei

Published: 2026-11-17T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Databases](<https://devfeed.tech/sources/arxiv-databases.md>)

Topics: [json lines format](<https://devfeed.tech/topics/json-lines-format.md>), [Vectors & LLMs](<https://devfeed.tech/topics/vectors-llms.md>), [Retrieval Augmented Generation (RAG)](<https://devfeed.tech/topics/retrieval-augmented-generation-rag.md>)

Tags: [algorithm](<https://devfeed.tech/tags/algorithm.md>), [complexity](<https://devfeed.tech/tags/complexity.md>), [cs-db](<https://devfeed.tech/tags/cs-db.md>), [cs-ds](<https://devfeed.tech/tags/cs-ds.md>), [cs-ir](<https://devfeed.tech/tags/cs-ir.md>), [encoding](<https://devfeed.tech/tags/encoding.md>), [geospatial](<https://devfeed.tech/tags/geospatial.md>), [index](<https://devfeed.tech/tags/index.md>), [json](<https://devfeed.tech/tags/json.md>), [language-models](<https://devfeed.tech/tags/language-models.md>), [large-language-models-llms](<https://devfeed.tech/tags/large-language-models-llms.md>), [llm-prompts](<https://devfeed.tech/tags/llm-prompts.md>), [llms](<https://devfeed.tech/tags/llms.md>), [openstreetmap](<https://devfeed.tech/tags/openstreetmap.md>), [pubchem](<https://devfeed.tech/tags/pubchem.md>), [query](<https://devfeed.tech/tags/query.md>), [rag](<https://devfeed.tech/tags/rag.md>), [reasoning](<https://devfeed.tech/tags/reasoning.md>), [retrieval](<https://devfeed.tech/tags/retrieval.md>), [retrieval-augmented-generation](<https://devfeed.tech/tags/retrieval-augmented-generation.md>), [retrieval-augmented-generation-rag](<https://devfeed.tech/tags/retrieval-augmented-generation-rag.md>), [structure](<https://devfeed.tech/tags/structure.md>), [structured](<https://devfeed.tech/tags/structured.md>)

### AI overview

jXBW enables direct substructure search over JSONL collections using a compressed index, avoiding the per-query structure reconstruction used by recent RAG systems. It merges repeated tree structures across objects, encodes them with the eXtended Burrows-Wheeler Transform (XBW), and searches in three phases. The authors say query cost depends on candidate matches and results rather than requiring a full collection scan. Across seven real-world datasets, including PubChem (1 million compounds) and OpenStreetMap (6.6 million objects), jXBW reportedly beat the strongest tree-based baseline by 16x on the smallest dataset and up to 2,800x on the largest. Saxon did not finish 1,000 PubChem queries within 24 hours; jXBW searches over million-object JSONL collections were reported in the sub-millisecond range.

### Source excerpt

arXiv:2508.12536v5 Announce Type: replace Abstract: Providing structured information to large language models (LLMs) has been shown to improve multi-step reasoning and factual accuracy (e.g., StructRAG, ICLR 2025; RAS, ICLR 2026); recent retrieval-augmented generation (RAG) systems reconstruct structure from retrieved text on every query. When the corpus is already structured, as in JSON Lines (JSONL), used for LLM prompts, chemical compounds, and geospatial objects, this per-query rebuilding can be replaced by direct structural retrieval. Its core operation is substructure search: finding all JSON objects in a collection that contain a given query tree. Existing approaches index each object separately, so index space and query time grow with collection size; XML-based engines add conversion overhead and semantic mismatches. We propose jXBW, a compressed index for fast substructure search over JSONL, combining three innovations: (i) a merged tree representation that consolidates repeated structures across objects, (ii) a succinct tree encoding based on the eXtended Burrows-Wheeler Transform (XBW), and (iii) a three-phase substructure search algorithm on this encoding. Together they achieve query-dependent complexity in compressed space: the search avoids a full scan of the collection, and its cost grows with the numbers of candidate matches and results. Experiments on seven real-world datasets, including PubChem (1 million compounds) and OpenStreetMap (6.6 million objects), show that jXBW outperforms the strongest tree-based baseline by 16x on the smallest dataset and by up to 2,800x on the largest, while the XML-based engine Saxon does not complete the 1,000-query workload on PubChem within 24 hours. jXBW brings substructure search over million-object JSONL collections into the sub-millisecond range.

## Estimating Outpatient Visit Prices from Hospital Price Transparency Data using Large Language Models

DevFeed: [Estimating Outpatient Visit Prices from Hospital Price Transparency Data using Large Language Models](<https://devfeed.tech/articles/estimating-outpatient-visit-prices-from-hospital-price-transparency-data-using-large-language-models-137554.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.20261>)

Author: Kurmanbek Kaiyrbekov, Nicholas J Dobbins, Sean D Mooney

Published: 2026-11-17T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>), [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Retrieval Augmented Generation (RAG)](<https://devfeed.tech/topics/retrieval-augmented-generation-rag.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>), [Vectors & LLMs](<https://devfeed.tech/topics/vectors-llms.md>), [LLM evaluation / benchmarking](<https://devfeed.tech/topics/llm-evaluation-benchmarking.md>), [Machine learning](<https://devfeed.tech/topics/machine-learning.md>)

Tags: [abstract](<https://devfeed.tech/tags/abstract.md>), [accessible](<https://devfeed.tech/tags/accessible.md>), [analyze](<https://devfeed.tech/tags/analyze.md>), [approach](<https://devfeed.tech/tags/approach.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [availability](<https://devfeed.tech/tags/availability.md>), [based](<https://devfeed.tech/tags/based.md>), [best](<https://devfeed.tech/tags/best.md>), [bm25](<https://devfeed.tech/tags/bm25.md>), [challenges](<https://devfeed.tech/tags/challenges.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cl](<https://devfeed.tech/tags/cs-cl.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [data](<https://devfeed.tech/tags/data.md>), [hospital](<https://devfeed.tech/tags/hospital.md>), [language](<https://devfeed.tech/tags/language.md>), [language-models](<https://devfeed.tech/tags/language-models.md>), [large](<https://devfeed.tech/tags/large.md>), [large-language-models](<https://devfeed.tech/tags/large-language-models.md>), [models](<https://devfeed.tech/tags/models.md>), [prices](<https://devfeed.tech/tags/prices.md>), [transparency](<https://devfeed.tech/tags/transparency.md>), [using](<https://devfeed.tech/tags/using.md>)

### AI overview

Using hospital price transparency files from three Maryland-area hospitals, the study tested retrieval-augmented generation (RAG) with BM25, OpenAI embeddings, and a GPT-BM25 hybrid against a no-RAG baseline. GPT-BM25 performed best for procedure price estimates: average MAPE was 19.44% for individual procedures and 4.90% for procedure groups, versus 147.98% and 82.82% without RAG. In synthetic symptom-based visit simulations, the system identified about half of reference procedures and lowered MAPE versus an LLM-only baseline at two hospitals, with comparable results at the third. Gains were smaller when procedures had to be inferred from clinical information. The authors cite inconsistent file formats, unexplained duplicate entries, and limited insurer-negotiated rate availability as practical constraints.

### Source excerpt

arXiv:2611.20261v1 Announce Type: cross Abstract: Objective: Hospital care prices in the United States have increased substantially, yet patients lack effective tools to access and compare costs. We aimed to develop and evaluate an LLM-based approach for improving access to and use of hospital price transparency (PT) data. Materials and Methods: To estimate visit prices, we used a retrieval-augmented generation (RAG) framework to analyze PT files from three major hospitals in Maryland and the surrounding region. We compared BM25, OpenAI embedding models, and a GPT-BM25 hybrid against a no-RAG baseline for cost estimation of individual procedures and procedure groups, then evaluated the best approach for symptom-based visit cost estimation using synthetic simulations. Results: Across hospitals, the GPT-BM25 retriever achieved the best performance, with average price MAPE of 19.44% for individual procedures and 4.90% for groups, compared with 147.98% and 82.82%, respectively, for the no-RAG baseline. In simulations, our framework correctly identified approximately 50% of reference procedures and reduced MAPE relative to the LLM-only baseline at two of the three hospitals, while achieving comparable performance at the third. Discussion: Our framework uses RAG to improve care price estimation in real-world settings. Compared with the no-RAG baseline, it substantially improved cost estimates for both individual procedures and procedure groups. Improvements were smaller in synthetic simulations, likely because inferring procedures from clinical information is more complex. Practical challenges remain, including heterogeneous PT file formats, unexplained duplicate entries, and limited availability of insurer-negotiated rates. Conclusion: LLMs that effectively integrate PT data show strong potential to provide patients with more accessible and accurate healthcare cost estimates.

## RSIMed: Scaling Medical Foundation Models beyond Limited Clinical Data via Retrieval-Grounded Self-Evolution

DevFeed: [RSIMed: Scaling Medical Foundation Models beyond Limited Clinical Data via Retrieval-Grounded Self-Evolution](<https://devfeed.tech/articles/rsimed-scaling-medical-foundation-models-beyond-limited-clinical-data-via-retrieval-grounded-self-evolution-137318.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.18646>)

Author: Wei Dai, Qianchu Liu, Guanghui Qin, Xing Han, Tristan Naumann, Sheng Zhang, Paul Pu Liang

Published: 2026-11-17T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>), [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [llm-reasoning](<https://devfeed.tech/topics/llm-reasoning.md>), [AI Foundation Models](<https://devfeed.tech/topics/ai-foundation-models.md>), [enhancing AI model accuracy](<https://devfeed.tech/topics/enhancing-ai-model-accuracy.md>), [Vectors & LLMs](<https://devfeed.tech/topics/vectors-llms.md>), [Advanced AI workflows](<https://devfeed.tech/topics/advanced-ai-workflows.md>)

Tags: [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [diagnosis](<https://devfeed.tech/tags/diagnosis.md>), [foundation-models](<https://devfeed.tech/tags/foundation-models.md>), [grounded](<https://devfeed.tech/tags/grounded.md>), [less](<https://devfeed.tech/tags/less.md>), [medical](<https://devfeed.tech/tags/medical.md>), [models](<https://devfeed.tech/tags/models.md>), [recursive-self-improvement](<https://devfeed.tech/tags/recursive-self-improvement.md>), [retrieval](<https://devfeed.tech/tags/retrieval.md>), [rl](<https://devfeed.tech/tags/rl.md>), [scaling](<https://devfeed.tech/tags/scaling.md>), [self-improvement](<https://devfeed.tech/tags/self-improvement.md>), [tasks](<https://devfeed.tech/tags/tasks.md>), [train](<https://devfeed.tech/tags/train.md>)

### AI overview

RSIMed trains medical diagnostic models through retrieval-grounded, evolving tasks: a frozen copy of the solver creates questions from retrieved evidence and targets topics and difficulty the solver still struggles with. On rare-disease diagnosis using MIMIC-IV admissions, RSIMed-27B reaches 0.516 accuracy and RSIMed-9B 0.488 without external models; the abstract reports accuracy scaling with training compute, while a static generated-task pool collapses. The best stated GPT-5.6 setting scores 0.399 at 38 times RSIMed-27B's inference cost, and tasks from four external teachers, including GPT-5.5, produce a less accurate diagnostician than the solver's own tasks. RSIMed-RL-27B, trained only on diagnosis, also raises its adjusted HealthBench Professional score from 0.291 to 0.303. These are the reported results in the supplied abstract.

### Source excerpt

arXiv:2611.18646v1 Announce Type: cross Abstract: Rare diseases are hard to diagnose, and the labeled clinical records needed to train models for them are few, private, and slow to extend, since a specialist must read every case. A model that writes its own training tasks could learn beyond these records, but no program can check a model-written diagnosis, and a wrong answer key corrupts training. To tackle this dilemma between data scarcity and difficulty of data annotations, we introduce RSIMed, Recursive Self-Improvement in Medicine, one of the first self-evolving pipelines in which a task proposer co-evolves in context with a diagnostic solver on real clinical records. Seeded by training admissions, a frozen copy of the solver writes each clinical question from retrieved medical evidence, so its answer rests on a source, and steers topics and difficulty toward what the solver still gets wrong. On rare-disease diagnosis from MIMIC-IV admissions, the accuracy of RSIMed-27B grows linearly with training compute, while a static pool of generated tasks collapses. Without any external model, RSIMed-27B and RSIMed-9B reach 0.516 and 0.488 diagnostic accuracy, above every off-the-shelf GPT-5.6 setting, whose best reaches 0.399 at 38 times the inference cost of RSIMed-27B. Tasks written by four external teachers, including GPT-5.5, train a less accurate diagnostician than the solver's own. Trained only on diagnosis, RSIMed-RL-27B also improves on the physician-written tasks of HealthBench Professional, from 0.291 to 0.303 in adjusted score.

## One Question, Four Voices: How Advice for Alzheimer's Caregiving Differs Between Caregivers, Clinicians, and Large Language Models

DevFeed: [One Question, Four Voices: How Advice for Alzheimer's Caregiving Differs Between Caregivers, Clinicians, and Large Language Models](<https://devfeed.tech/articles/one-question-four-voices-how-advice-for-alzheimer-s-caregiving-differs-between-caregivers-clinicians-and-large-language-models-137300.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.18604>)

Author: Congning Ni, Jinkyung Katie Park, Yang Li, Sarvech Qadir, Sean S. Huang, James S. Powers, Julia A. Hiner, Vania Leung, Patricia Commiskey, Lijun Song, Qingxia Chen, Pamela J. Wisniewski, Bradley A. Ma

Published: 2026-11-17T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [Large Language Model](<https://devfeed.tech/topics/llm.md>), [AI Chat](<https://devfeed.tech/topics/ai-chat.md>), [Vectors & LLMs](<https://devfeed.tech/topics/vectors-llms.md>), [Language Engineering](<https://devfeed.tech/topics/language-engineering.md>)

Tags: [advice](<https://devfeed.tech/tags/advice.md>), [ai](<https://devfeed.tech/tags/ai.md>), [chatgpt](<https://devfeed.tech/tags/chatgpt.md>), [community](<https://devfeed.tech/tags/community.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cl](<https://devfeed.tech/tags/cs-cl.md>), [cs-hc](<https://devfeed.tech/tags/cs-hc.md>), [emotion](<https://devfeed.tech/tags/emotion.md>), [financial](<https://devfeed.tech/tags/financial.md>), [guidance](<https://devfeed.tech/tags/guidance.md>), [hybrid](<https://devfeed.tech/tags/hybrid.md>), [language-models](<https://devfeed.tech/tags/language-models.md>), [large-language-models](<https://devfeed.tech/tags/large-language-models.md>), [legal](<https://devfeed.tech/tags/legal.md>), [llm](<https://devfeed.tech/tags/llm.md>), [llms](<https://devfeed.tech/tags/llms.md>), [modeling](<https://devfeed.tech/tags/modeling.md>), [online](<https://devfeed.tech/tags/online.md>), [peer](<https://devfeed.tech/tags/peer.md>), [retrieval](<https://devfeed.tech/tags/retrieval.md>), [scope](<https://devfeed.tech/tags/scope.md>), [structured](<https://devfeed.tech/tags/structured.md>), [style](<https://devfeed.tech/tags/style.md>)

### AI overview

The study compared answers to 85 real Alzheimer's and related dementia caregiving questions from peer caregivers, physicians, ChatGPT, and retrieval-augmented CareGPT. Peers offered time-anchored narratives and broader community-grounded coverage, including emotional, legal, financial, and daily-care concerns; clinicians gave concise, clinically bounded advice; and both LLMs produced formal, verbose, structured responses concentrated on advice. LLMs scored higher than peers on politeness and emotional-support markers and were comparable to clinicians on informativeness and physician alignment, but were less narrative and contextually situated. CareGPT differed little from ChatGPT, suggesting its retrieval and prompt injection did not substantially change its communicative stance. The authors conclude AI may complement, but not replace, human caregiver support.

### Source excerpt

arXiv:2611.18604v1 Announce Type: cross Abstract: Online caregiver forums are an important source of emotional and informational support for Alzheimer's disease and related dementias (ADRD). We compare responses to 85 real caregiver questions from ALZConnected across four responder groups: peer caregivers, physicians, ChatGPT, and a retrieval-augmented CareGPT grounded in peer discussions. Using 28 measures spanning linguistic form, emotional support, informational support, and topic scope, we examine how these voices differ when answering the same questions. Peers provided time-anchored narratives and community grounding; clinicians offered concise, advice-forward, clinically bounded guidance; and LLMs produced a formal, verbose, highly structured consultation style. CareGPT showed minimal measurable separation from ChatGPT, suggesting that retrieval plus prompt injection alone was insufficient to shift communicative stance. LLM outputs scored higher than peers on politeness and emotion-support markers and were comparable to clinicians on informativeness and physician-alignment measures, but were less narrative and less contextually situated. Topic modeling showed that LLM responses concentrated on advice-oriented themes, whereas peers covered a broader range, including emotional processing, legal and financial concerns, and daily care. These findings suggest that AI can complement, but not replace, human contributions to ADRD caregiver support and motivate hybrid systems that combine structured AI delivery with the distinct strengths of human voices.

## Samyama: A Unified Graph-Vector Database with In-Database Optimization, Agentic Enrichment, and Hardware Acceleration

DevFeed: [Samyama: A Unified Graph-Vector Database with In-Database Optimization, Agentic Enrichment, and Hardware Acceleration](<https://devfeed.tech/articles/samyama-a-unified-graph-vector-database-with-in-database-optimization-agentic-enrichment-and-hardware-acceleration-135433.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2603.08036>)

Author: Madhulatha Mandarapu, Sandeep Kunkunuru

Published: 2026-11-16T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Databases](<https://devfeed.tech/sources/arxiv-databases.md>)

Topics: [graph-database](<https://devfeed.tech/topics/graph-database.md>), [Vectors & LLMs](<https://devfeed.tech/topics/vectors-llms.md>), [database-optimization](<https://devfeed.tech/topics/database-optimization.md>), [RDF (Resource Description Framework)](<https://devfeed.tech/topics/rdf.md>), [Agentic AI](<https://devfeed.tech/topics/what-is-agentic-ai.md>), [RAPIDS](<https://devfeed.tech/topics/rapids.md>)

Tags: [22](<https://devfeed.tech/tags/22.md>), [acceleration](<https://devfeed.tech/tags/acceleration.md>), [agentic](<https://devfeed.tech/tags/agentic.md>), [analytics](<https://devfeed.tech/tags/analytics.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [cloud](<https://devfeed.tech/tags/cloud.md>), [cpu](<https://devfeed.tech/tags/cpu.md>), [cs-db](<https://devfeed.tech/tags/cs-db.md>), [cuda](<https://devfeed.tech/tags/cuda.md>), [cypher](<https://devfeed.tech/tags/cypher.md>), [database](<https://devfeed.tech/tags/database.md>), [database-optimization](<https://devfeed.tech/tags/database-optimization.md>), [databases](<https://devfeed.tech/tags/databases.md>), [edge](<https://devfeed.tech/tags/edge.md>), [enrichment](<https://devfeed.tech/tags/enrichment.md>), [graph](<https://devfeed.tech/tags/graph.md>), [hardware](<https://devfeed.tech/tags/hardware.md>), [hardware-acceleration](<https://devfeed.tech/tags/hardware-acceleration.md>), [optimization](<https://devfeed.tech/tags/optimization.md>), [vector](<https://devfeed.tech/tags/vector.md>), [vector-database](<https://devfeed.tech/tags/vector-database.md>)

### AI overview

Samyama combines graph and vector storage, analytics, RDF/SPARQL, 22 query-callable metaheuristic solvers, and LLM-based graph enrichment in a Rust engine. At v1.8.0, tests on rented Linux hosts report 123K-152K nodes/s ingestion and PageRank iterations of 6.7 ms at 10,000 nodes and 236 ms at one million on a 16-vCPU instance. On an NVIDIA A16, CUDA matches CPU performance at 10,000 nodes and reaches 1.62x and 2.70x at 100,000 and one million, respectively, with 1.1-1.35 seconds of first-call context initialization. The abstract withdraws earlier GPU, Cypher-throughput, index-scaling, and vector-search figures as unreproducible. It reports 99.9% of evaluated openCypher TCK scenarios passing; in a same-host comparison with Neo4j 5 and FalkorDB, all 14 complex reads were within 5x of the best competitor and 7 were faster.

### Source excerpt

arXiv:2603.08036v4 Announce Type: replace Abstract: Modern data architectures fragment across graph databases, vector stores, analytics engines and optimization solvers, forcing ETL between them. We present Samyama, a graph-vector database in Rust that unifies these workloads in one engine: a RocksDB-backed store with MVCC, a vectorized executor, a cost-based planner, a CSR analytics engine, RDF and SPARQL, 22 metaheuristic solvers callable from the query language, HNSW vector indexing, and agentic enrichment that expands a graph via LLMs. It has been run to billion-edge scale: 74.3M nodes and 1.07B edges from four biomedical sources on one machine for $2.50 of spot compute. This version re-measures the system at release v1.8.0 on rented Linux hosts a reader can boot, replacing earlier figures taken on a Mac Mini that nobody could re-run. On a 16-vCPU cloud instance, ingestion reaches 123K-152K nodes/s and PageRank costs 6.7 ms per iteration at ten thousand nodes and 236 ms at a million. The previously reported 8.2x GPU speedup is withdrawn; in its place we measure the CUDA path, which no earlier version measured at all, on a rented NVIDIA A16: parity with the CPU at ten thousand nodes, 1.62x at a hundred thousand and 2.70x at a million, with 1.1-1.35 s of context initialisation on the first call. The earlier Cypher-throughput figure, its claim of near-constant index-driven scaling, and a vector-search figure are withdrawn as unreproducible. openCypher conformance, once estimated at ~90%, is now measured at 99.9% of evaluated TCK scenarios. Against Neo4j 5 and FalkorDB on one host over an identical SF10 extract, 14 of 14 SNB complex reads land within 5x of the best competitor and 7 of 14 are faster.

## StreamEmbed: Real-Time Indexing from Pixels for Screenshot Retrieval

DevFeed: [StreamEmbed: Real-Time Indexing from Pixels for Screenshot Retrieval](<https://devfeed.tech/articles/streamembed-real-time-indexing-from-pixels-for-screenshot-retrieval-135913.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.18516>)

Author: Dayoon Ko, Yichuan Wang, Soyeong Jeong, Joseph E. Gonzalez, Gunhee Kim, Sewon Min

Published: 2026-11-16T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [screenshot](<https://devfeed.tech/topics/screenshot.md>), [real-time](<https://devfeed.tech/topics/real-time.md>), [search-engines](<https://devfeed.tech/topics/search-engines.md>), [Cache](<https://devfeed.tech/topics/cache.md>), [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [Vectors & LLMs](<https://devfeed.tech/topics/vectors-llms.md>)

Tags: [adapter](<https://devfeed.tech/tags/adapter.md>), [agent](<https://devfeed.tech/tags/agent.md>), [agents](<https://devfeed.tech/tags/agents.md>), [benchmark](<https://devfeed.tech/tags/benchmark.md>), [blocks](<https://devfeed.tech/tags/blocks.md>), [computer-use](<https://devfeed.tech/tags/computer-use.md>), [cs-cv](<https://devfeed.tech/tags/cs-cv.md>), [cs-ir](<https://devfeed.tech/tags/cs-ir.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [embedding](<https://devfeed.tech/tags/embedding.md>), [embeddings](<https://devfeed.tech/tags/embeddings.md>), [encoding](<https://devfeed.tech/tags/encoding.md>), [exploits](<https://devfeed.tech/tags/exploits.md>), [fast](<https://devfeed.tech/tags/fast.md>), [faster](<https://devfeed.tech/tags/faster.md>), [foundation](<https://devfeed.tech/tags/foundation.md>), [frontier](<https://devfeed.tech/tags/frontier.md>), [gpu](<https://devfeed.tech/tags/gpu.md>), [indexing](<https://devfeed.tech/tags/indexing.md>), [pixels](<https://devfeed.tech/tags/pixels.md>), [real-time](<https://devfeed.tech/tags/real-time.md>), [retrieval](<https://devfeed.tech/tags/retrieval.md>), [screenshot](<https://devfeed.tech/tags/screenshot.md>)

### AI overview

StreamEmbed speeds up visual indexing of incoming screenshots by exploiting repeated pixels: it skips spatial blocks that duplicate a neighbor and reuses cached keys and values for blocks unchanged between screenshots. A learned adapter helps preserve fidelity to embeddings computed from full screenshots. On ScreenMemEval, with 10,426 questions over 184 hours of histories, it encoded up to 3.7x faster than full-screenshot encoding on an M2 Pro GPU without sacrificing accuracy, and improved Recall@10 by up to 6.3 points over OCR indexing at matched latency. Its vector index used 0.13 GB of RAM for a simulated year of screenshots. These are reported benchmark results; the supplied text does not describe evaluation beyond that setup.

### Source excerpt

arXiv:2611.18516v1 Announce Type: cross Abstract: Screen history can serve as searchable visual memory for people and agents, but unlike an offline corpus, it must be encoded and indexed as each screenshot arrives. OCR-based indexing is fast but loses visual information; visual encoding takes over two seconds. Yet screens change incrementally: most pixels repeat within and across screenshots. This raises a question: do visual encoders need to recompute their representations from scratch for every new screenshot? We present StreamEmbed, a system for incremental visual embedding that exploits this redundancy. Guided solely by pixel differences, it combines spatial token pruning, which skips blocks that repeat a neighbor, with temporal KV reuse, which reuses cached keys and values for blocks unchanged across screenshots. A small learned adapter compensates for pruning and reuse to preserve fidelity to full-screenshot embeddings. On ScreenMemEval, our benchmark of 10,426 questions over 184 hours of screen histories from two humans and a computer-use agent, StreamEmbed sets a new accuracy-latency Pareto frontier on an M2 Pro GPU: it encodes up to 3.7x faster than full-screenshot encoding without sacrificing accuracy and improves accuracy over OCR-based indexing by up to 6.3 Recall@10 points at matched latency. A memory-efficient vector index stores a simulated year of screenshots in 0.13 GB of RAM. Together, these results make real-time indexing of screen history in its native visual form practical, laying a foundation for fully local visual assistants.

## How to Evaluate Snowflake Cortex Embedding Models for Enterprise Search

DevFeed: [How to Evaluate Snowflake Cortex Embedding Models for Enterprise Search](<https://devfeed.tech/articles/how-to-evaluate-snowflake-cortex-embedding-models-for-enterprise-search-133909.md>)

Original publisher: [Read original article](<https://hackernoon.com/how-to-evaluate-snowflake-cortex-embedding-models-for-enterprise-search?source=rss>)

Author: Akshay Sharma

Published: 2026-11-14T03:45:02Z

Content type: article

Language: en

Sources: [HackerNoon](<https://devfeed.tech/sources/hackernoon.md>)

Topics: [Embeddings](<https://devfeed.tech/topics/embeddings.md>), [hybrid-search](<https://devfeed.tech/topics/hybrid-search.md>), [AI-driven information retrieval](<https://devfeed.tech/topics/ai-driven-information-retrieval.md>), [search-engines](<https://devfeed.tech/topics/search-engines.md>), [Vectors & LLMs](<https://devfeed.tech/topics/vectors-llms.md>), [AI Chat](<https://devfeed.tech/topics/ai-chat.md>)

Tags: [8k](<https://devfeed.tech/tags/8k.md>), [account](<https://devfeed.tech/tags/account.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-assistant](<https://devfeed.tech/tags/ai-assistant.md>), [arctic](<https://devfeed.tech/tags/arctic.md>), [cortex](<https://devfeed.tech/tags/cortex.md>), [cost](<https://devfeed.tech/tags/cost.md>), [data-engineering](<https://devfeed.tech/tags/data-engineering.md>), [dimensions](<https://devfeed.tech/tags/dimensions.md>), [embedding](<https://devfeed.tech/tags/embedding.md>), [embedding-models](<https://devfeed.tech/tags/embedding-models.md>), [enterprise](<https://devfeed.tech/tags/enterprise.md>), [enterprise-ai](<https://devfeed.tech/tags/enterprise-ai.md>), [enterprise-search](<https://devfeed.tech/tags/enterprise-search.md>), [models](<https://devfeed.tech/tags/models.md>), [performance](<https://devfeed.tech/tags/performance.md>), [rag](<https://devfeed.tech/tags/rag.md>), [retrieval](<https://devfeed.tech/tags/retrieval.md>), [search](<https://devfeed.tech/tags/search.md>), [semantic-search](<https://devfeed.tech/tags/semantic-search.md>), [snowflake](<https://devfeed.tech/tags/snowflake.md>), [snowflake-cortex](<https://devfeed.tech/tags/snowflake-cortex.md>), [text](<https://devfeed.tech/tags/text.md>), [vector](<https://devfeed.tech/tags/vector.md>), [vector-databases](<https://devfeed.tech/tags/vector-databases.md>)

### AI overview

Snowflake Cortex makes embedding generation through AI_EMBED straightforward, but choosing among its eight text models requires testing retrieval on the enterprise workload. The article recommends starting with a compact English model for short English documents, evaluating multilingual candidates on cross-language queries, and comparing long-context models with chunking for lengthy or mixed-topic documents. Larger vectors and longer context windows do not guarantee better retrieval; dimensions alone are not a quality measure. Semantic similarity can miss exact error codes and identifiers, so hybrid lexical and vector retrieval may help. Evaluate candidates on representative queries and relevant documents, weigh failure impact, check account and regional availability and current pricing, and record model and chunking versions because changing models may require re-embedding the corpus. No comparative benchmark winner is claimed.

### Source excerpt

Snowflake Cortex offers eight text embedding models, but which one should you choose? Explore vector dimensions, multilingual retrieval, performance, and cost.

## VecDBLens: An Evaluatology-Based Framework for Vector Database Diagnosis

DevFeed: [VecDBLens: An Evaluatology-Based Framework for Vector Database Diagnosis](<https://devfeed.tech/articles/vecdblens-an-evaluatology-based-framework-for-vector-database-diagnosis-132691.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.15105>)

Author: Yutong Zhou, Guoxin Kang, Wenxin Zhou, Haoyu He, Yuedong Zhu, Lei Wang, Wanling Gao, Fanda Fan, Jianfeng Zhan

Published: 2026-11-13T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Databases](<https://devfeed.tech/sources/arxiv-databases.md>)

Topics: [Vectors & LLMs](<https://devfeed.tech/topics/vectors-llms.md>), [Embeddings](<https://devfeed.tech/topics/embeddings.md>), [Performance Testing](<https://devfeed.tech/topics/performance-testing.md>)

Tags: [abstract](<https://devfeed.tech/tags/abstract.md>), [all](<https://devfeed.tech/tags/all.md>), [analysis](<https://devfeed.tech/tags/analysis.md>), [applications](<https://devfeed.tech/tags/applications.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [based](<https://devfeed.tech/tags/based.md>), [benchmarking](<https://devfeed.tech/tags/benchmarking.md>), [bounded](<https://devfeed.tech/tags/bounded.md>), [can](<https://devfeed.tech/tags/can.md>), [cause](<https://devfeed.tech/tags/cause.md>), [challenge](<https://devfeed.tech/tags/challenge.md>), [comparisons](<https://devfeed.tech/tags/comparisons.md>), [cost](<https://devfeed.tech/tags/cost.md>), [cs-db](<https://devfeed.tech/tags/cs-db.md>), [database](<https://devfeed.tech/tags/database.md>), [database-benchmarking](<https://devfeed.tech/tags/database-benchmarking.md>), [databases](<https://devfeed.tech/tags/databases.md>), [diagnosis](<https://devfeed.tech/tags/diagnosis.md>), [framework](<https://devfeed.tech/tags/framework.md>), [vector](<https://devfeed.tech/tags/vector.md>), [vector-database](<https://devfeed.tech/tags/vector-database.md>)

### AI overview

VecDBLens diagnoses vector database performance by comparing one retrieval function at a time under matched dataset, workload, parameter, and runtime conditions, then summarizing results across datasets and workloads. It uses ANOVA to examine interactions among functions, aiming to make benchmarking reproducible and identify whether representation, indexing, or query execution drives performance. The abstract reports that embedding quality upper-bounds retrieval performance, so apparent index failures may stem from representation limits; it also finds that overly aggressive index settings can alter neighborhood topology and reduce recall. The available text is an abstract, so it does not provide implementation details or quantify these effects.

### Source excerpt

arXiv:2611.15105v1 Announce Type: new Abstract: Vector databases are increasingly deployed as retrieval pipelines in which representation, indexing, and query-time execution jointly determine accuracy, throughput, and cost. Existing vector database evaluations primarily focus on indexing while treating embeddings as fixed inputs and ignoring the role of query-time execution, making it difficult to decide whether a bottleneck should be addressed by changing the embedding model, the index, or the query-processing functions. To address the above limitation, we introduce VecDBLens, an Evaluatology-based framework that treats the function under diagnosis as an evaluated object and all remaining dataset, workload, parameter, and runtime factors as matched evaluation conditions. VecDBLens attributes performance differences to individual evaluated objects through comparisons under matched evaluation conditions, summarizing their performance and stability across multiple datasets and workloads. It further analyzes interactions among objects using analysis of variance (ANOVA), thereby turning vector database benchmarking from leaderboard ranking into a reproducible diagnostic procedure. VecDBLens reveals two key insights. First, retrieval performance is fundamentally upper-bounded by embedding quality, revealing that many apparent indexing failures are therefore representation-side limitations. Second, stronger index parameterization is not always beneficial: overly aggressive configurations can reshape neighborhood topology and cause recall regression. These results challenge the common practice of evaluating vector databases primarily through indexing and show that real-world limitations arise from constraints across retrieval functions rather than any single function alone. VecDBLens provides a principled basis for identifying the true performance drivers of vector databases and for guiding their design in practical applications.

## Pointing the Way, Hiding the Destination: Practical Private Dense Retrieval at Scale

DevFeed: [Pointing the Way, Hiding the Destination: Practical Private Dense Retrieval at Scale](<https://devfeed.tech/articles/pointing-the-way-hiding-the-destination-practical-private-dense-retrieval-at-scale-131963.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2608.25735>)

Author: Peichun Hua, Danyang Chen, Junan Zhang, Haifeng Sun, Jingyu Wang, Diwen Xue, Mingyu Li, Yunming Xiao

Published: 2026-11-13T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>), [arXiv: Cryptography and Security](<https://devfeed.tech/sources/arxiv-cryptography-and-security.md>)

Topics: [information retrieval](<https://devfeed.tech/topics/information-retrieval.md>), [Cryptography](<https://devfeed.tech/topics/cryptography.md>), [AI-driven information retrieval](<https://devfeed.tech/topics/ai-driven-information-retrieval.md>), [Vectors & LLMs](<https://devfeed.tech/topics/vectors-llms.md>), [hashing](<https://devfeed.tech/topics/hashing.md>)

Tags: [abstract](<https://devfeed.tech/tags/abstract.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [can](<https://devfeed.tech/tags/can.md>), [candidate](<https://devfeed.tech/tags/candidate.md>), [circuits](<https://devfeed.tech/tags/circuits.md>), [clusters](<https://devfeed.tech/tags/clusters.md>), [code](<https://devfeed.tech/tags/code.md>), [cryptographic](<https://devfeed.tech/tags/cryptographic.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cr](<https://devfeed.tech/tags/cs-cr.md>), [cs-ir](<https://devfeed.tech/tags/cs-ir.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [deep](<https://devfeed.tech/tags/deep.md>), [differential-privacy](<https://devfeed.tech/tags/differential-privacy.md>), [efficiency](<https://devfeed.tech/tags/efficiency.md>), [either](<https://devfeed.tech/tags/either.md>), [embedding](<https://devfeed.tech/tags/embedding.md>), [encrypted](<https://devfeed.tech/tags/encrypted.md>), [existing](<https://devfeed.tech/tags/existing.md>), [filter](<https://devfeed.tech/tags/filter.md>), [practical](<https://devfeed.tech/tags/practical.md>), [private](<https://devfeed.tech/tags/private.md>), [retrieval](<https://devfeed.tech/tags/retrieval.md>), [scale](<https://devfeed.tech/tags/scale.md>)

### AI overview

The paper uses learned deep hashing to make private dense retrieval more efficient: a randomized binary code selects a short candidate list, then encrypted reranking and oblivious key transfer protect the query and final selection. Across five zero-shot corpora of 25K to 5.4M documents, lists of 200-500 candidates closely match full-corpus retrieval. On 2.68M NQ passages over a 10-Gbps link, the protocol adds 0.38 seconds (6.5%) to a 128-token Qwen3-32B RAG pipeline. The released code satisfies directional metric differential privacy and substantially reduces embedding-inversion and property-inference leakage.

### Source excerpt

arXiv:2608.25735v3 Announce Type: replace-cross Abstract: Hosted retrieval-augmented generation (RAG) and semantic search allow users to query valuable provider-held corpora, raising two competing demands: to hide each query and chosen result, yet reveal only the documents that the user is authorized to receive. Existing cryptographic approaches either make this costly by processing the entire corpus for every query, or sacrifice quality for efficiency by scanning a few clusters. We repurpose learned deep hashing as a private filter: a randomized binary code points the provider to a short candidate list, while encrypted reranking and oblivious key transfer protect the precise query and final selection. This shortlist short-circuits full-corpus cryptographic search without sacrificing retrieval quality: with 200-500 candidates, it closely matches full-corpus retrieval across five zero-shot corpora spanning 25K to 5.4M documents. On the full 2.68M-passage NQ corpus over a 10-Gbps link, our protocol only adds 0.38 seconds, or 6.5%, to a 128-token Qwen3-32B RAG pipeline. The released code satisfies directional metric differential privacy (DP) and substantially reduces embedding-inversion and property-inference leakage, demonstrating that a carefully learned shortlist can make private dense retrieval both accurate and practical.

## Automatic Knowledge Graph Construction and Query for Earthquake Catalogs

DevFeed: [Automatic Knowledge Graph Construction and Query for Earthquake Catalogs](<https://devfeed.tech/articles/automatic-knowledge-graph-construction-and-query-for-earthquake-catalogs-131953.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2607.24984>)

Author: Yuxin Zhou, Huai Zhang, S. Mostafa Mousavi

Published: 2026-11-13T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [Vectors & LLMs](<https://devfeed.tech/topics/vectors-llms.md>), [база знаний](<https://devfeed.tech/topics/tag-c51cc4bf76e5.md>), [Enhancing AI model performance](<https://devfeed.tech/topics/enhancing-ai-model-performance.md>)

Tags: [catalog](<https://devfeed.tech/tags/catalog.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [graph](<https://devfeed.tech/tags/graph.md>), [graphrag](<https://devfeed.tech/tags/graphrag.md>), [knowledge](<https://devfeed.tech/tags/knowledge.md>), [knowledge-graph](<https://devfeed.tech/tags/knowledge-graph.md>), [physics-geo-ph](<https://devfeed.tech/tags/physics-geo-ph.md>), [prompt](<https://devfeed.tech/tags/prompt.md>), [prompting](<https://devfeed.tech/tags/prompting.md>), [query](<https://devfeed.tech/tags/query.md>), [rag](<https://devfeed.tech/tags/rag.md>), [temporal](<https://devfeed.tech/tags/temporal.md>)

### AI overview

The paper applies GraphRAG directly to raw tabular earthquake catalogs to build queryable knowledge graphs without manual data structuring. It evaluates the approach on a reservoir-adjacent swarm, the 2019 Ridgecrest sequence, and the 2021 Maduo Mw7.4 aftershock sequence, checking results against catalog-derived ground truth and a rule-based reference graph. Four seismology-informed prompt fixes reportedly eliminated targeted fabrications and improved mechanism reasoning. Compared with a vector RAG baseline, the graph approach adds catalog-wide summarization and temporal-stage comparison. The abstract also flags two pitfalls but does not specify them, so the evidence here does not establish their nature or the extent of the evaluation.

### Source excerpt

arXiv:2607.24984v2 Announce Type: replace-cross Abstract: In recent years, the number of events in earthquake catalogs has significantly increased due to the utilization of more effective deep learning based detectors and phase pickers but answering open ended questions such as what characterizes this sequence? remains constrained by rigid spatiotemporal windowing and subjective expert interpretation. We present the first systematic application of graph based retrieval augmented generation GraphRAG directly to raw, tabular catalog records across three independently featured catalogs, a reservoir adjacent swarm, the 2019 Ridgecrest tectonic sequence, and the 2021 Maduo Mw7.4 aftershock sequence. Without the need for manual data structuring, the pipeline builds structurally complete, queryable knowledge graphs for all three. Rigorous evaluation individually verified against catalog derived ground truth and a rule based reference graph exposes failure modes, and four seismology informed prompt fixes eliminate all targeted fabrications while sharply improving mechanism reasoning. A vector RAG baseline demonstrates the graph layers distinctive value, catalog wide summarization and temporal stage comparison. In addition, we have identified two main pitfalls that need attention. GraphRAG thus offers a practical, transferable, near zero cost query interface for earthquake catalogs, where careful prompting ensures the results are consistently accurate and trustworthy.

## Calibrated Early Termination with Bounded Optimality Gaps for Constrained Vector Search

DevFeed: [Calibrated Early Termination with Bounded Optimality Gaps for Constrained Vector Search](<https://devfeed.tech/articles/calibrated-early-termination-with-bounded-optimality-gaps-for-constrained-vector-search-132699.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.16148>)

Author: Xinyu Zhang, Jingze Li, Baihua Zheng, Weiwei Sun

Published: 2026-11-13T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Databases](<https://devfeed.tech/sources/arxiv-databases.md>), [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [ann](<https://devfeed.tech/topics/ann.md>), [Embeddings](<https://devfeed.tech/topics/embeddings.md>), [Vectors & LLMs](<https://devfeed.tech/topics/vectors-llms.md>)

Tags: [cost](<https://devfeed.tech/tags/cost.md>), [cs-db](<https://devfeed.tech/tags/cs-db.md>), [cs-ir](<https://devfeed.tech/tags/cs-ir.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [datasets](<https://devfeed.tech/tags/datasets.md>), [duality](<https://devfeed.tech/tags/duality.md>), [dynamic-programming](<https://devfeed.tech/tags/dynamic-programming.md>), [embeddings](<https://devfeed.tech/tags/embeddings.md>), [failure](<https://devfeed.tech/tags/failure.md>), [faster](<https://devfeed.tech/tags/faster.md>), [graph](<https://devfeed.tech/tags/graph.md>), [indexes](<https://devfeed.tech/tags/indexes.md>), [linear](<https://devfeed.tech/tags/linear.md>), [lower-bound](<https://devfeed.tech/tags/lower-bound.md>), [multiple-testing](<https://devfeed.tech/tags/multiple-testing.md>), [nearest-neighbor](<https://devfeed.tech/tags/nearest-neighbor.md>), [queries](<https://devfeed.tech/tags/queries.md>), [query](<https://devfeed.tech/tags/query.md>), [recall](<https://devfeed.tech/tags/recall.md>), [recommendation](<https://devfeed.tech/tags/recommendation.md>), [search](<https://devfeed.tech/tags/search.md>), [vector](<https://devfeed.tech/tags/vector.md>), [vector-search](<https://devfeed.tech/tags/vector-search.md>)

### AI overview

CaST calibrates early stopping for constrained vector search to bound the probability of returning an inexact answer by ε, with confidence 1−δ. It restricts stopping rules to those monotone in declared observations, computes a nested chain of staircase rules by dynamic programming, and uses fixed-sequence binomial tests to calibrate them. Lagrangian weak duality supplies a lower bound on the best cost within that class across risk levels without a union bound over rules. In replay on two million-item datasets, CaST cost at most 1.132 times the class optimum for risk levels 0.03-0.10, with 95% joint confidence per setting; on held-out queries at a common risk target, other calibrated rules tested took 4.8%-45.5% more mean time. The guarantee and optimality comparison are limited to the declared rule class and reported experiments.

### Source excerpt

arXiv:2611.16148v1 Announce Type: new Abstract: Vector search over learned embeddings is the retrieval layer of recommendation and retrieval-augmented generation, and many applications cap how many answer items one category may contribute. Graph indexes answer a query by best-first traversal, and where it stops determines its cost and whether its answer is exact. We study early termination for such constrained $k$-nearest-neighbor ($k$-NN) queries, with plain $k$-NN as a special case, and seek the cheapest stopping rule whose answer is exact with probability at least $1-\varepsilon$. Learned early-termination methods instead target average recall, which does not guarantee exact answers. Multiple testing can calibrate any parameterized stopping rule to our target, but it cannot tell how much a better rule with the same risk and observations would save. Answering this needs a lower bound on the optimum of a rule family too large to enumerate, valid for the query population. Our key idea is to declare the observations a rule may use and to consider every rule monotone in them. Risk-constrained stopping then becomes a linear program over an order polytope, whose Lagrangian minimizers form a nested chain of staircase rules computed exactly by dynamic programming. Fixed-sequence binomial tests along this chain calibrate our rule, CaST, whose failure probability is at most $\varepsilon$ with confidence $1-\delta$. Lagrangian weak duality bounds the optimum of the declared class from below at all risk levels at once, without a union bound over the rules. On two datasets of one million items each, CaST replayed on logged searches costs at most 1.132 times the class optimum for risk levels from 0.03 to 0.10, at 95% joint confidence per setting. Larger savings thus need a rule outside the class. At a common risk target on held-out queries, CaST is faster than every other calibrated rule we time, which take 4.8% to 45.5% more mean time.

## A 23M-Parameter Model Still Wins: 338 Embedding Models Ranked by Real Downloads

DevFeed: [A 23M-Parameter Model Still Wins: 338 Embedding Models Ranked by Real Downloads](<https://devfeed.tech/articles/a-23m-parameter-model-still-wins-338-embedding-models-ranked-by-real-downloads-130592.md>)

Original publisher: [Read original article](<https://hackernoon.com/a-23m-parameter-model-still-wins-338-embedding-models-ranked-by-real-downloads?source=rss>)

Author: Writings, Papers and Blogs on Text Models

Published: 2026-11-12T16:50:13Z

Content type: article

Language: en

Sources: [HackerNoon](<https://devfeed.tech/sources/hackernoon.md>)

Topics: [sentence-transformers](<https://devfeed.tech/topics/sentence-transformers.md>), [Vectors & LLMs](<https://devfeed.tech/topics/vectors-llms.md>), [hugging face](<https://devfeed.tech/topics/hugging-face.md>), [qwen](<https://devfeed.tech/topics/qwen.md>), [Qdrant](<https://devfeed.tech/topics/qdrant.md>), [Ollama](<https://devfeed.tech/topics/ollama.md>)

Tags: [2026](<https://devfeed.tech/tags/2026.md>), [alibaba](<https://devfeed.tech/tags/alibaba.md>), [alibaba-qwen](<https://devfeed.tech/tags/alibaba-qwen.md>), [allen-institute-for-ai](<https://devfeed.tech/tags/allen-institute-for-ai.md>), [amazon](<https://devfeed.tech/tags/amazon.md>), [amazon-bedrock](<https://devfeed.tech/tags/amazon-bedrock.md>), [amazon-titan](<https://devfeed.tech/tags/amazon-titan.md>), [apache](<https://devfeed.tech/tags/apache.md>), [apache-2-0](<https://devfeed.tech/tags/apache-2-0.md>), [api](<https://devfeed.tech/tags/api.md>), [downloads](<https://devfeed.tech/tags/downloads.md>), [embedding](<https://devfeed.tech/tags/embedding.md>), [embedding-models](<https://devfeed.tech/tags/embedding-models.md>), [embedding-models-ranked](<https://devfeed.tech/tags/embedding-models-ranked.md>), [embedding-text-models](<https://devfeed.tech/tags/embedding-text-models.md>), [embeddings](<https://devfeed.tech/tags/embeddings.md>), [hackernoon-top-story](<https://devfeed.tech/tags/hackernoon-top-story.md>), [hugging-face](<https://devfeed.tech/tags/hugging-face.md>), [model](<https://devfeed.tech/tags/model.md>), [models](<https://devfeed.tech/tags/models.md>), [open-source-embedding-models](<https://devfeed.tech/tags/open-source-embedding-models.md>), [openai](<https://devfeed.tech/tags/openai.md>), [parameter](<https://devfeed.tech/tags/parameter.md>), [qwen](<https://devfeed.tech/tags/qwen.md>), [rag](<https://devfeed.tech/tags/rag.md>), [textmodels-ranked-by-downloads](<https://devfeed.tech/tags/textmodels-ranked-by-downloads.md>)

### AI overview

The directory's November 11, 2026 snapshot ranks 213 Hugging Face open-weight embedding models by verified 30-day download counts where available. all-MiniLM-L6-v2 (23M parameters) leads with 177M downloads--38% of the tracked total and more than the next four models combined--while the top 10 account for 74%. The article contrasts this adoption ranking with benchmark-based lists: small, established models lead, multilingual models are prominent, Qwen3-Embedding is a newer high-ranking family, and Qdrant conversions illustrate database-provided builds. Hosted APIs lack public download counts and are grouped alphabetically by provider; some listed models are retired. Ollama models are ordered by size because Ollama publishes no counts. The article suggests English defaults (all-MiniLM-L6-v2 or bge-small-en-v1.5), multilingual options (bge-m3 or multilingual-e5), and Qwen3-Embedding for top quality when GPU resources permit, while advising readers to test on their own data.

### Source excerpt

all-MiniLM-L6-v2 got 177M Hugging Face downloads in 30 days. See all 338 embedding models ranked by real downloads, from BAAI and Qwen to OpenAI.

## RAG Retrieval Lost 40 Points When Distracting Passages Grew to 2.68M

DevFeed: [RAG Retrieval Lost 40 Points When Distracting Passages Grew to 2.68M](<https://devfeed.tech/articles/rag-retrieval-lost-40-points-when-distracting-passages-grew-to-2-68m-130836.md>)

Original publisher: [Read original article](<https://towardsdatascience.com/rag-retrieval-lost-40-points-when-distracting-passages-grew-to-268m/>)

Author: Abdullahi Dattijo

Published: 2026-11-12T16:30:00Z

Content type: article

Language: en

Sources: [Towards Data Science](<https://devfeed.tech/sources/towards-data-science.md>)

Topics: [Retrieval Augmented Generation (RAG)](<https://devfeed.tech/topics/retrieval-augmented-generation-rag.md>), [information retrieval](<https://devfeed.tech/topics/information-retrieval.md>), [search-engines](<https://devfeed.tech/topics/search-engines.md>), [Embeddings](<https://devfeed.tech/topics/embeddings.md>), [context window](<https://devfeed.tech/topics/context-window.md>), [Vectors & LLMs](<https://devfeed.tech/topics/vectors-llms.md>)

Tags: [bm25](<https://devfeed.tech/tags/bm25.md>), [context](<https://devfeed.tech/tags/context.md>), [embeddings](<https://devfeed.tech/tags/embeddings.md>), [information-retrieval](<https://devfeed.tech/tags/information-retrieval.md>), [large-language-models](<https://devfeed.tech/tags/large-language-models.md>), [llm](<https://devfeed.tech/tags/llm.md>), [llm-evaluation](<https://devfeed.tech/tags/llm-evaluation.md>), [position](<https://devfeed.tech/tags/position.md>), [rag](<https://devfeed.tech/tags/rag.md>), [retrieval](<https://devfeed.tech/tags/retrieval.md>), [search](<https://devfeed.tech/tags/search.md>), [test](<https://devfeed.tech/tags/test.md>)

### AI overview

BM25 retrieval was nearly unchanged as a collection grew to 1,801 arXiv papers, but on 2.68 million Wikipedia passages its success at 10 fell about 40 points versus a 10,000-passage baseline; the tests differed in several ways, so the author cannot identify which difference drove the gap. In a smaller Wikipedia test, adding 20 unsupported passages from the answer's own page reduced retrieval success from 90-92% to 81-84%. The added passages often ranked highly because BM25 doubled title-match weight, not because they were semantically similar; embeddings were not tested there. With Qwen3 4B, answers scored worse when the judged passage was lower or absent in the retrieved context, and placing it first made scores nearly equal across conditions. The author recommends testing with questions written by people unfamiliar with the documents and measuring the answer passage's position. Results are limited by positive-only relevance labels, unusually high judged-passage density in the smallest Wikipedia collection, and a single reviewer for the same-page labels.

### Source excerpt

A reproducible test of BM25, embeddings, and parent documents shows why author-written evals overrate search, and why answer position in context matters. The post RAG Retrieval Lost 40 Points When Distracting Passages Grew to 2.68M appeared first on Towards Data Science.

[Next page](<https://devfeed.tech/topics/vectors-llms.md?cursor=WyIyMDI2LTExLTEyVDE2OjMwOjAwKzAwOjAwIiwgIjViNzA3OWU2LTVjYmUtNDM1YS1iODZiLTljNWY5N2NmNWEyYiJd>)