Retrieval pipelines, vector search, chunking and reranking: how models are grounded in a corpus instead of their weights.
The article explains that Retrieval-Augmented Generation (RAG) is a retrieval system, while agents are responsible for action. The author built a distinct layer that explicitly connects retrieval to action, and tested this setup across nine tasks alongside standalone RAG and agent systems.
SemMSA introduces a latent semantic‑aided framework for multimodal sentiment analysis that leverages large language models to generate rich sentiment‑relevant semantics. The method employs Cross‑modal Semantic Refinement (CSR) to fuse visual, acoustic, and language features in a frozen LLM embedding space, and Cross‑modal Spectral Alignment (CSA) to align these refined semantics with all modalities via spectral enhancement of kernel Gram matrices. Experiments on SIMS, MOSI, and MOSEI benchmarks show that SemMSA achieves state‑of‑the‑art performance.
TWIST is a new benchmark suite designed to evaluate the quality of interventions in conversational memory systems, focusing on how well these systems act correctly at belief change points. It includes four tracks—tension detection, draft vetting, belief-consistent answering, and sensitive recall governance—each paired with hard-negative controls to prevent gaming. The benchmark is rigorously validated through double annotation, adjudication, and a separability audit, revealing that current models face trade-offs between contradiction recall and false-positive rates.
The paper investigates when it is better to return an existing draft answer or revise it using retrieved evidence in retrieval‑augmented QA systems. By grading both the draft and its candidate revision with the same correctness judge, the authors define a paired effect called recoverability and train policies to predict it before revision. Experiments on 25,870 open‑domain questions show that a recoverability‑based scorer outperforms a draft‑correctness scorer across multiple Llama setups, improving accuracy–revision trade‑offs and closing a significant portion of the oracle gap, though it still applies harmful revisions in a substantial fraction of cases.
The paper introduces ELCD, a latent conflict detector that verifies LLM outputs after generation to catch instruction conflicts that static input checks miss. ELCD builds a hidden-state representation from the final-token embedding and the mean-pooled response embedding, then trains a pairwise margin ranking objective to distinguish compliant from drifting responses. Experiments on five large language models show ELCD outperforms baselines, boosting PR-AUC for Llama‑2‑7B by ~30 percentage points and cutting FPR95 for Mistral‑7B to 2.67%.
The paper introduces Domain Recentering with Confidence Calibration (DRC), a training‑free technique that adapts CLIP to unlabeled target images by fitting a Gaussian mixture and subtracting a posterior‑weighted average of component means from each embedding. It further corrects residual class bias using a log‑prior adjustment based on confidence‑weighted predictions. DRC outperforms other methods, raising average accuracy on cross‑domain datasets by 4.13 and 5.07 points over zero‑shot CLIP for ViT‑B/16 and ResNet‑50, and maintains gains under ImageNet distribution shifts.
CRISS is a retrieval‑augmented generation chatbot designed to aid cancer registrars by providing rapid, citation‑supported answers to complex coding and staging questions. It uses a domain‑specific knowledge base of national registry standards, indexed as dense embeddings, to retrieve relevant passages that ground responses generated by a large language model. In evaluations, RAG configurations consistently outperformed non‑RAG baselines across easy, medium, and hard questions, achieving higher grounding and semantic similarity scores while maintaining human oversight for final decisions.
The paper introduces the Support Vector Graph (SVG), a graph index for vector search that uses kernel methods to guarantee navigability in both metric and non‑metric vector spaces, such as inner product similarity. It shows that popular indices like HNSW and DiskANN are special cases of SVG, and proposes SVG‑L0, which adds an ℓ₀ sparsity constraint to enforce bounded out‑degree while maintaining computational efficiency.
The paper investigates how trajectory fine‑tuning can enhance small language models (SLMs) as next‑action controllers in retrieval‑augmented question answering. By building a seven‑way action‑prediction task from teacher search traces, the authors fine‑tune SLMs and cross‑lingual SLMs (xSLMs) using LoRA and evaluate on 1,646 held‑out examples, achieving a macro‑F1 of 0.6536 with Granite 4.1 3B. In an end‑to‑end controller/generator swap experiment on 149 trajectories, the fine‑tuned model improves Exact Match from 0.7530 to 0.7946 and token F1 from 0.7783 to 0.8295, demonstrating that trajectory supervision boosts action prediction and evidence‑recording behavior.
The paper introduces a modular, multi-level framework for diagnosing malignant melanoma that incorporates contextual information from the lesion, patient, and population levels. It features an anatomic-site aware masked transformer that models patient context by considering all lesions and their locations, and uses learnable demographics embeddings to capture population statistics. Experiments on the SIIM-ISIC 2020 dataset show that adding lesion context and metadata improves specificity by 17.15% and 7.14% respectively, while also enhancing balanced accuracy.
META (Memory Enhanced Trading Agent) is a new agent-based trading framework that augments large language models with episodic memory. It combines specialized indicator agents—such as Trend, MACD, Stochastic, RSI, SMA, AVWAP, and Heikin‑Ashi—with a Decision Agent that fuses their reports, while a Memory module retrieves and updates past trading episodes encoded as market state embeddings with outcomes and reflections. By recalling relevant experiences and adaptively reweighting signals under similar market regimes, META improves directional accuracy and robustness in short‑horizon evaluations, offering regime‑aware, interpretable, and low‑latency decision‑making for financial trading.
ASIRF (Agentic Sensitive Information Redaction Framework) is a system that retrieves domain‑specific definitions of sensitive information from a flexible knowledge base at inference time, eliminating the need for retraining when adapting to new domains. It offers two architectures—a three‑call multi‑agent pipeline and a single‑agent variant—and has been evaluated on ten small open‑weight models across eight datasets, including out‑of‑distribution fictional domains. In 68 of 80 model‑domain combinations (85 %), ASIRF’s recall surpasses that of the OpenAI Privacy Filter, with most shortfalls limited to the filter’s training‑distribution domains.
RD‑JEPA is a joint‑embedding predictive architecture designed for self‑supervised pretraining on reaction‑diffusion trajectories. The model is pretrained on five parameterized systems and then adapted to three held‑out systems that were not seen during pretraining. Using as few as one, five, or ten complete trajectories from a held‑out system, RD‑JEPA outperforms five supervised surrogate baselines, an independently trained control that removes the trajectory‑dependent predictive latent pathway, and an architecture‑matched model trained from scratch, achieving lower mean relative discrete β field error and mean absolute spatial first‑difference error across various output resolutions, forecast horizons, and adaptation trajectory choices.
The paper investigates whether reasoning always benefits universal multimodal embeddings (UMEs). By comparing the discriminative and reasoning-driven branches of UME-R1, the authors find that while reasoning improves positive similarity in 56.6% of cases, it also creates 15.7% false-helpful instances where hard negatives are drawn closer. Diagnostic analyses reveal that reasoning often de‑condenses retrieved neighborhoods and that chain‑of‑thought tokens encode evidence common to both positives and hard negatives. Based on these insights, the authors introduce SURE, a utility router that boosts UME-R1‑7B by 1.5 points and consistently improves other embedding models on MMEB‑V2 without retraining or extra VLM passes.
The paper introduces a pipeline that localizes frames from historical PTZ maritime video onto a reference panorama and then selects a context‑aware, diverse subset for ship detection training. By combining SuperPoint‑LightGlue localization, weather and solar‑state metadata, and diversity sampling, the method reduces 40,718 candidate frames to just 220 images for annotation. Fine‑tuned YOLOv26 on this compact set achieves high detection performance (AP50 ≈ 94.8%) while cutting annotation effort by 99.5%.
ChunkRank is an open‑source Python library that automatically determines chunk boundaries based on a target model’s tokenizer and context window, and then selects an answer from independently produced chunk candidates. It includes a registry of 90 models from 15 providers and six answer‑selection methods, and requires only three core dependencies. Experiments show that token‑exact budgeting is important across 11 languages, and that for several datasets no content‑based ranker outperforms simply taking the first non‑empty answer due to reader abstention on chunks lacking the answer.
ReCalMatch introduces a reliability‑calibrated semantic framework for semi‑supervised fine‑grained visual recognition, addressing the problem of overconfident pseudo‑label errors that arise when visually similar categories produce incorrect high‑confidence predictions. The method constructs class‑conditioned semantic prototypes from class names and domain‑specific aspects, and computes a visual‑semantic agreement score to calibrate pseudo‑label reliability alongside prediction confidence and entropy. Experiments on datasets such as CUB‑200‑2011, Stanford Dogs, NABirds, and iNaturalist18 demonstrate that ReCalMatch consistently improves strong SSL baselines, especially in low‑label regimes where pseudo‑label noise is most severe.
The paper presents a retrieval‑augmented generation pipeline for answering regulatory compliance questions in finance. It builds a three‑stage retriever on LegalBERT and a compact 2B–12B generator served with 4‑bit quantization, achieving a Recall@10 of 0.774 on the ObliQA benchmark and improving answer quality via RAFT‑LoRA fine‑tuning. However, the adapted models fail to transfer to Australian case‑law questions, and a closed‑book model performs almost as well while lacking verifiable grounding.
The paper introduces VeriSpeak, a benchmark of 3,879 spoken claims for evaluating fact verification in Large Audio Language Models (LALMs). It shows a clear modality gap: models that verify written claims well often fail on spoken versions, and retrieval alone offers limited improvement. Combining retrieval with explicit reasoning yields the best performance, reaching 86.1% accuracy and demonstrating the need for grounded reasoning over retrieved evidence in speech misinformation detection.
DeGRe is a dense‑supervised generative reranking framework designed to improve multi‑stage recommender systems by addressing label bias and credit assignment issues. It uses an offline Lookahead Evaluator with beam search to generate dense supervision signals, which are distilled into a lightweight Online Generator that can perform efficient greedy decoding at inference time. Experiments show that DeGRe outperforms baselines on public benchmarks and industrial datasets, and it has been successfully deployed on Taobao Flash Shopping to enhance online recommendations.