Retrieval pipelines, vector search, chunking and reranking: how models are grounded in a corpus instead of their weights.
arXiv:2610.23589v1 Announce Type: new Abstract: Backdoor attacks have emerged as a critical security threat in machine learning, where adversaries embed triggers within poisoned training samples to m...
arXiv:2610.23730v1 Announce Type: new Abstract: Recent deep facial phenotyping methods can support the recognition of genetic syndromes, but patient images are often scarce and unevenly distributed a...
arXiv:2610.22630v1 Announce Type: cross Abstract: Traffic agencies need current and near-future speeds on every link of their network, but their fixed sensors cover only part of it. We study joint no...
arXiv:2610.22885v1 Announce Type: cross Abstract: Classifier-free guidance (CFG) is a standard component of conditional diffusion models, and guided generation performance has become the \textit{de f...
arXiv:2609.31868v2 Announce Type: replace Abstract: Model-predictive control with Joint-Embedding Predictive Architectures (JEPAs) provides a strong zero-shot goal-reaching planner, but it is only ef...
arXiv:2610.22428v1 Announce Type: new Abstract: In a small language model the vocabulary is the single largest part. In Gemma 3 270M the embedding table holds 168M of 268M parameters, and because qua...
arXiv:2610.23205v1 Announce Type: new Abstract: Large Language Models (LLMs) remain unreliable for extremely low-resource languages whose writing systems conflict with high-resource pretraining prior...
The paper introduces SYNTRA, a self‑improving framework that transforms a small set of retrieval examples into reusable synthesis instructions for generating training data. By inferring implicit information needs and relevance criteria from an unlabelled corpus and refining instructions through feedback, SYNTRA produces effective training data across 24 datasets and 15 domains, improving retrieval performance even for strong embedding models. Experiments show that SYNTRA’s self‑improvement yields better results than human calibration on four tasks and scales effectively, with continued gains on MS MARCO across a 32‑fold range of synthetic data sizes.
PocketTSLM is a mobile-native deployment of the OpenTSLM SoftPrompt time‑series language model that runs entirely on a consumer smartphone. By injecting encoded time‑series windows as embeddings instead of tokenizing them as text, PocketTSLM preserves temporal structure while using far fewer tokens and staying within the device’s context and memory limits. The model has been benchmarked on 30‑second EEG sleep‑staging clips and 10‑second ECG recordings, achieving on‑device inference with no health data transmitted to the cloud.
The paper introduces a training‑free, target‑aware refinement method for 3D Gaussian language maps distilled from vision‑language features. By blending the visual embedding within user‑specified target regions with the corresponding text embedding using a convex combination, the method improves target localization while preserving open‑vocabulary queryability. The authors derive the optimal blending coefficient analytically, avoid hyperparameters, and demonstrate consistent gains across three benchmarks, including a real‑world robot navigation demo.
The paper examines whether transcripts of model solutions can reveal capabilities after benchmark scores saturate. By simulating saturation with historical data, the authors test protocols that use only correctly solved transcripts to predict performance on unsaturated items. Pairwise judging of transcripts outperforms other methods, reducing prediction error by 5.3×, though its effectiveness can vary across distribution shifts.
The paper presents a lightweight, scalable workflow for refining ontologies, demonstrated on the CASTON casting-domain ontology. It combines a fine‑tuned ontology embedding model to generate candidate axioms and replacement options with a Large Language Model (LLM) that proposes repair suggestions. In a small expert evaluation, the default setup—using a CASTON‑fine‑tuned embedding model and Qwen3‑8B—yielded high‑confidence repairs that were all accepted by domain experts.
The paper presents a privacy‑preserving text inference framework using CKKS homomorphic encryption for health‑misinformation classification and semantic retrieval. It combines BioBERT, MPNet, and TF‑IDF features for classification, achieving an accuracy of 0.971 and macro‑F1 of 0.950, while a late‑fusion similarity model reaches a Pearson correlation of 0.8691. Optimized ciphertext retrieval reduces latency from 222.028 s to 5.168 s and communication overhead from 6.11 GB to 2.13 MB over 7,128 items.
CIM-Embed is a framework that learns instruction‑conditioned multimodal embeddings for clustering. It uses supervised contrastive learning with same‑group positives and negatives from other groups, and handles tasks with varying group counts via heterogeneous structured batches. At inference, the model maps an object’s image, text, and instruction to an embedding, and HDBSCAN clusters these embeddings without requiring a pre‑specified cluster count. The authors introduce ProductMIC, a benchmark of 5,373 Amazon products and 449 grouping tasks, and report a mean per‑task V‑measure of 95.01, outperforming the best general‑purpose baseline.
CARA is a training‑free, causally guided retrieval agent designed for question answering over text‑rich knowledge graphs. It evaluates retrieved evidence using two causal criteria—consequentiality and essentiality—alongside path consistency, and aggregates results from multiple agents via reciprocal rank fusion. Experiments on three STaRK datasets show that CARA outperforms existing retrievers, improving answer quality and faithfulness while examining fewer graph nodes and requiring fewer retrieval operations.
The paper presents a multi‑source Retrieval‑Augmented Generation (RAG) AI tutor designed to help non‑native English speaker (NNES) students in computing courses. It can process questions in foreign languages—currently Mandarin—and give bilingual answers, generate glossaries of technical terms, and translate queries. In tests on 91 Piazza posts, the tutor answered 83.52 % correctly, translated 76.92 % of questions, and produced glossaries for 64.47 % of them.
The paper introduces a lightweight privacy-preserving technique for Retrieval-Augmented Generation (RAG) systems that uses random projection to transform both datastore and query embeddings. This approach maintains semantic similarity for effective retrieval while significantly reducing the risk of data extraction attacks. Experiments across various RAG architectures and datasets demonstrate that the method outperforms prior differential privacy-based techniques in both retrieval and generation performance while offering comparable empirical privacy.
CoTS: Competitive Token Selection for Latent Reasoning in Large Language Models proposes a training‑free framework that replaces binary latent reasoning decisions with continuous token‑level allocation. By using the probability competition between the two most likely tokens, CoTS dynamically balances discrete commitment and soft exploration at each reasoning step, eliminating the need for manually tuned thresholds or additional forward passes. Experiments on six reasoning benchmarks show that CoTS consistently outperforms standard Chain‑of‑Thought and other state‑of‑the‑art training‑free latent reasoning methods.
The paper introduces Loud and Clear (LoC), a method that enhances multimodal document retrieval-augmented generation by fully utilizing visual token capacity and refining retrieved pages to highlight evidence. LoC combines higher-resolution pages with OCR of query-relevant regions, achieving significant accuracy gains over eight prior methods on multiple benchmarks. The study demonstrates that the way retrieved pages are presented to the generator is a critical bottleneck beyond retrieval itself.
The paper investigates privacy risks of language embeddings used in recommender systems. It demonstrates that a simple cosine‑similarity attack can recover item identities with high accuracy when an adversary has auxiliary text, even when the descriptions are independently sourced. The study also shows that such leakage can lead to user re‑identification from sparse ratings and evaluates obfuscated descriptions as a mitigation, revealing a trade‑off between privacy and utility.