MeddeX is a medical report analyzer that fuses a QLoRA‑fine‑tuned LLaMA‑3 (8B) model with Retrieval‑Augmented Generation and a Gemini Vision layer to produce concise, evidence‑backed summaries of medical documents. It processes text‑native PDFs, scanned PDFs, and radiological images via a hybrid extraction pipeline featuring a garbage‑text detection mechanism for corrupted or UTF‑16‑encoded files. The system adds report‑scoped retrieval, an automated longitudinal comparison engine, and a self‑healing backfill pipeline, and is backed by a FastAPI backend, Supabase vector store, and Next.js frontend. Evaluation shows significant gains over the base LLaMA‑3‑Med42‑8B model across ROUGE, BLEU, and BERTScore metrics, underscoring the value of domain‑specific fine‑tuning.
By Rajdeep Pandhare, Rohit Kundaikar, Yakub Kojar, Shreyansh Tiwari
The paper investigates how the positional geometry of cached key‑value (KV) states affects inference quality in sparse KV compression. It introduces RoPERelay, a method that reassigns compact, order‑preserving RoPE coordinates and re‑rotates cached key phases to better align with current queries, without requiring new selection or extra cache budget. Experiments on two models and multiple selectors show that RoPERelay improves answer quality in most cases, notably raising Llama‑3.1‑8B‑Instruct accuracy from 23.6% to 35.4% on LongMemEval‑S with only a 1–3% runtime overhead.
By Wanyang Hong, Zhaoning Zhang
The paper investigates why large language models (LLMs) miss evidence in multi‑hop retrieval‑augmented generation (RAG). It introduces ExecutionRAG, a training‑free register that tracks open sub‑questions and their supporting evidence, allowing the authors to distinguish between true retrieval failures and search drift. Using this method on 1,000 MuSiQue questions, the study shows a significant reduction in missed evidence but also reveals that other error types, such as premature resolutions and abstentions, increase, shifting the bottleneck from search guidance to evidence verification.
By Jiaqi Wei, Wenxuan Huang, Chengdong Xu, Xiang Zhang, Wanli Ouyang
The paper demonstrates that large language models (LLMs) used to preliminarily assess medical device documentation can produce inconsistent results across different commercial deployments and settings. By constructing a fully deterministic pipeline—including OCR, chunking, embedding, indexing, retrieval, and inference—the authors achieved byte‑identical outputs across repeated runs, varied machines, and even interruptions. The study shows that end‑to‑end determinism is possible and provides a verifiable foundation for audit and change control in regulatory compliance.
By Ryosuke Suzuki
The paper introduces Temporal State Tracking Memory (TEM), a training‑free framework that stores facts as versioned records linked to their sources and retrieves the relevant versions in source order, labeling historical records before passing them to the answer model. TEM incorporates event memory and original dialogue turns to fill gaps in structured facts. Evaluations on 420 questions from 600 sessions with 60 personas show high key‑entity hit rates (75.71–97.62%) and significant improvements when historical labels are added, while keeping evidence compact (94–118 tokens per answer).
By Pengfeng Song, Yunkai Zhuang, Liye Shi
The paper presents a specialized NMT model for the Masewalkopa variant of Nawatl, a low‑resource language from Mexico. Using a 70,000‑pair multi‑variant foundation, the authors curate 2,665 parallel pairs and augment them with 192 elder‑transcribed sentences via asymmetric back‑translation. A custom finite‑state analyzer reduces unanalyzed forms from 82.0% to 19.9% and supplies functional syntax tags as prefixes, enabling the model to reuse subword embeddings and achieve a dev chrF++ of 58.46 and test chrF++ scores around 58.0. The final LoRA‑v9c‑b model is quantized with CTranslate2 int8, cutting CPU latency from 137.5 to 1.8–2.6 seconds per request for cost‑bounded Cloud Run deployment.
By Luis A. Mercado-Campos, Alexis Palmer
VELD-QA is a large-scale benchmark for visual evidence grounding and question answering in long documents, comprising 15,000 human-authored question-answer pairs across 5,859 PDFs with 372,445 pages. The dataset includes detailed annotations of supporting pages, visual evidence elements, and descriptions of cited visual content, enabling separate evaluation of retrieval, localization, and interpretation. Experiments on nine vision-language models and various retrieval-augmented systems reveal significant gaps in evidence localization and visual interpretation, with oracle accuracy peaking at 81.7% and enrichment of visual-element annotations narrowing performance gaps by 17–24%.
By Zhivar Sourati, Marianne Menglin Liu, Rongguang Wang, Zheng Wang, Yazhe Hu, Lan Guo, Yuying Wang, Mengqing Guo, Bhavesh Dnyaneshwar Gore, Shubham Pathak, Ramapriya Ranganath, Sandeep Sharma, Tao Sheng, Sujith Ravi, Dan Roth
The paper introduces LoopMol, a closed‑loop molecular inverse‑design framework that uses a large language model (LLM) as a surrogate to reason directly over task instructions, SMILES optimization histories, and oracle feedback. At each iteration, the LLM selects informative reference molecules and optionally a concise guidance sentence, conditioning the next generation round and producing an inspectable optimization trace. Experiments on six drug and material tasks across three Llama‑based backbones show that the LLM‑based variant outperforms one‑shot generation and competes with or surpasses Gaussian‑process Bayesian optimization, with performance varying by task type.
By Yaoyao Xu, Xinjian Zhao, Xiaozhuang Song, Lei Bai, Tianshu Yu
The paper introduces CE‑RAG4EM, a cost‑efficient framework that applies retrieval‑augmented generation (RAG) to entity matching (EM) by using blocking‑based batch retrieval and generation. It demonstrates that, in a zero‑shot setting, CE‑RAG4EM achieves comparable or better matching quality than naive RAG while significantly reducing end‑to‑end overhead. Experiments on nine real‑world EM benchmarks show that the approach closes the knowledge grounding gap for LLM‑based EM, especially on datasets with ambiguous or sparse attributes, and provides clear guidance on balancing accuracy and computational cost through blocking size and retrieval granularity.
By Chuangtao Ma, Zeyu Zhang, Arijit Khan, Sebastian Schelter, Paul Groth
FuseAdapt is a parameter‑efficient framework for multimodal semantic segmentation that performs fusion within the adaptation space of a frozen vision foundation model (VFM). Instead of a separate fusion network, each modality receives a lightweight adaptation branch, and only low‑rank residual activations are routed across modalities inside the frozen backbone. The method also introduces Predictive Modality Modeling (PMM) to handle missing modalities during training, achieving significant mIoU gains on three benchmarks.
By Xin Zhang, Robby T. Tan
EOS-Story is a training‑free framework that improves visual storytelling by balancing identity consistency with scene‑specific adaptation. It separates narrative prompts into identity and scene sub‑prompts and introduces Cross‑Contextual Bridging to inject appearance cues into the scene embedding. In the vision space, Adaptive Identity Preservation aligns identity features to the scene scale and uses Identity‑Aware Self‑Attention to avoid structural artifacts, achieving competitive results on the extended Consistory+ benchmark.
By SeungJu Cha, Ye-Chan Kim, Kwanyoung Lee, Dong-Jin Kim
The paper evaluates whether multimodal large language models (MLLMs) can serve as forensic detectors of deepfake facial images. Six MLLMs—including GPT-4o, GPT-5, GPT-5.2, Gemini 2.5, Gemini 3, and Seed 1.6—were tested on a dataset of 400 images (50 authentic and 350 synthetic) generated by seven leading image models. Results show that several MLLMs achieve high detection accuracy, with GPT-5.2 reaching 97.9% in pairwise comparisons, but the presence of invisible SynthID watermarks can reduce detection performance by about 10% for Gemini.
By Olivia Salamon, Virginia N. L. Franqueira, Shujun Li
WaveLLM introduces a wavelet-domain entropy model that leverages large language models for lossless image compression. By decomposing images into multi-level wavelet subbands, it transforms heterogeneous prediction tasks into structured, statistically homogeneous subproblems, enabling more effective autoregressive probability estimation. Using a dedicated numerical interface and task-adaptive prompting, WaveLLM, built on Llama‑3.2‑3B with LoRA, achieves state‑of‑the‑art bpsp results on Kodak, CLIC.pro, and DIV2K, outperforming both PixelCNN‑based codecs and recent LLM‑based compressors.
By Yue Yu, Bowei Kang, Changsheng Gao, Dong Liu, Xiaoyan Sun
As assistants move from reactive answering toward proactive intervention, memory changes role: it no longer only supplies the answer to a query, it determines whether and when to speak at all. Prior w...
Retrieval is a fundamental task in multimodal systems: given an image or text query, the goal is to find the most relevant images or texts in a collection, often using a shared embedding space. Reliab...
The paper introduces a dual‑view approach for time‑series language models, combining line‑plot visualizations that preserve absolute measurements with normalized delay embeddings that remove amplitude scaling and offsets while exposing temporal structure. By fusing these complementary representations through visual tokens and multi‑level feature fusion, the model achieves significant performance gains on several benchmarks, improving TSRBench accuracy from 44.5% to 53.2% and outperforming dedicated TSLM baselines. The results demonstrate that leveraging both absolute and relative temporal cues enables more effective reasoning over time‑series data.
EOS-Story is a training‑free framework that improves visual storytelling by balancing identity consistency with scene‑specific adaptation. It separates narrative prompts into identity and scene sub‑prompts, then uses Cross‑Contextual Bridging to inject appearance cues into the scene embedding. In the vision space, Adaptive Identity Preservation aligns identity features to the scene scale and employs Identity‑Aware Self‑Attention to avoid structural artifacts, achieving strong results on the Consistory+_+ benchmark.
ARC‑SORT is a real‑time, training‑free multi‑object tracker designed for fast, direction‑changing athletes and dynamic camera motion in sports broadcasts. It uses an Interacting Multiple Model Kalman filter to adaptively size search regions based on per‑track uncertainty, and applies a single corrective identity swap after track crossings when appearance evidence indicates a mistake. The method achieves state‑of‑the‑art HOTA scores on SportsMOT and DanceTrack while running at ~50 fps on a laptop GPU.
Conditional memory expands language-model capacity through selective lookup, but the full embedding table must remain available for every possible input. We introduce Conditional Program (CP), a compr...
Language embeddings are increasingly used in recommenders to represent item content, but their privacy risks remain poorly understood. We study whether item embeddings can reveal item identities when...