Retrieval pipelines, vector search, chunking and reranking: how models are grounded in a corpus instead of their weights.
arXiv:2608.25727v2 Announce Type: replace Abstract: Large language models (LLMs) have recently advanced graph neural networks (GNNs) by enriching node representations with semantic information, givin...
arXiv:2610.04850v2 Announce Type: replace Abstract: Protein binding prediction is central to target identification, therapeutic binder design, and large scale screening, yet remains challenging becau...
arXiv:2610.10508v1 Announce Type: new Abstract: Prompted embedding models have recently received increasing attention, particularly for retrieval, where detailed retrieval instructions are provided a...
MorphCL introduces a self‑supervised pretraining framework for inertial‑based human activity recognition that explicitly models global motion structure through morphology‑aware grouping. By leveraging motion primitives and domain‑specific descriptors, MorphCL improves linear probing and fine‑tuning performance of encoders by up to 15 percentage points in F1‑score. Compared with existing foundation models, MorphCL‑pretrained encoders achieve comparable or better linear probing results while using 4,600× less data, and the learned embeddings exhibit clear, morphologically meaningful clusters that better separate kinematically similar activities.
arXiv:2610.09412v1 Announce Type: cross Abstract: Retrieval diversification is widely available in retrieval-augmented generation (RAG) frameworks, yet prior studies disagree on whether it improves r...
arXiv:2512.08617v2 Announce Type: replace Abstract: This tutorial focused on Healthcare Domain Applications of NLP, what we have achieved around HealthcareNLP, and the challenges that lie ahead for t...
arXiv:2604.05087v4 Announce Type: replace Abstract: Generative large language models (LLMs) are increasingly used as inference-time components in retrieval pipelines, for tasks such as query rewritin...
The paper introduces Shared Gaussianization (SG), a characteristic‑function test that evaluates Gaussianity of the average of two normalized views scaled by an independent χ_d radius. SG detects both misalignment and non‑uniformity in contrastive learning, vanishing exactly at aligned, uniform minimizers of population InfoNCE, and provides a sharp, dimension‑free bound on InfoNCE excess. The study also explores how SG behaves under different kernels, alignment weights, batch sizes, and latent‑variable models, revealing optimization gaps and the impact of nuisance channels.
arXiv:2607.04576v2 Announce Type: replace Abstract: LLM agents now often answer questions from knowledge bases they help maintain. A common intuition says progressive disclosure should make this chea...
arXiv:2610.08954v1 Announce Type: new Abstract: Video large language models (Vid-LLMs) excel at diverse video-language tasks by reasoning over selected frames. However, frame selection for long video...
arXiv:2610.09449v1 Announce Type: new Abstract: Zero-shot Chinese character recognition (ZS-CCR) aims to recognize characters whose categories are never observed during training, and typically relies...
arXiv:2610.09702v1 Announce Type: new Abstract: Ordinary prompt-based editing can cause latent watermark detection to fail without explicitly targeting the watermark. We benchmark eight watermark met...
arXiv:2610.09940v1 Announce Type: cross Abstract: Joint-embedding predictive architectures (JEPAs) predict masked or future observations in representation space, offering a natural source of predicti...
ORCA (Orthogonal Residual Compositional Alignment) is a method that improves text-to-image diffusion models by aligning the diffusion transformer’s latent space with a low‑rank target derived from a frozen visual encoder. It introduces an auxiliary loss that uses a predictor with an orthogonal basis parameterised by a learned residual between T5 and CLIP embeddings, providing a prompt‑dependent signal for selecting the visual readout subspace. Experiments on three diffusion‑transformer backbones show that ORCA improves FID and GenEval scores, especially on attribute binding, spatial relations, and multi‑object prompts, without adding inference‑time cost.
arXiv:2603. 13357v2 Announce Type: replace-cross Abstract: Bi-CamoDiffusion is introduced, an evolution of the CamoDiffusion framework for camouflaged object detection.
HySPE introduces a hyperbolic symplectic positional encoding that uses dual shears to create a conformally symplectic contraction with two spectral decay rates per channel pair. It eliminates exponential drift by diagonalizing the operator and applying blockwise coordinate rebasing, ensuring length‑independent numerical bounds while matching the forward latency of RoPE. Experiments on TinyShakespeare and a 51M‑parameter Transformer show HySPE maintains stable perplexity at long sequence lengths, outperforming RoPE in extrapolation scenarios.
arXiv:2607.28627v2 Announce Type: replace-cross Abstract: Long visual contexts challenge vision-language models: performance degrades as the number of distractors grows, and processing all tokens at...
The paper introduces AirCPU, an extraction‑oriented approach to over‑the‑air computation that uses joint source‑channel coding and a multi‑layer nested lattice architecture to create a structured wireless superposition. Unlike traditional embedding‑oriented designs, AirCPU operates directly on continuous device data, avoiding separate quantization and enabling progressive resolution through hierarchically scaled components transmitted over a bounded digital constellation. The authors formalize decoupled resolution, showing that with low decoding error probability, channel noise and constellation limits have negligible distortion, and they extend the method to fading MACs with collective, successive, and direct computation mechanisms to broaden the reliable operating regime.
EngramEdit introduces a method for decoupled knowledge updates in large language models using conditional memory architectures. By computing target memory representations that align with updated facts across multiple expressions, it jointly adjusts shared n‑gram embeddings while penalizing frequent ones to preserve unrelated knowledge. Experiments demonstrate near‑perfect editing success, improved multi‑hop reasoning, and strong performance under chain‑of‑thought prompting, all while maintaining general capabilities.
arXiv:2610.09375v1 Announce Type: new Abstract: Organizations increasingly use frontier language models to analyze customer feedback, but answer quality also depends on how that feedback is organized...