「text」の検索結果
49 件Deployed language models are updated continually, and regulators increasingly accept changes that were declare
Diffusion models represent a powerful class of generative models known for their solid theoretical foundations
Individual-level trajectories are essential for applications that rely on user-level mobility behavior. Existi
Parallel decoding speeds up masked diffusion language models by writing several tokens per forward pass. Publi
Separating touching nuclei requires local boundary detail, while cell type classification requires broader mor
Recent work has improved computer-use agents (CUAs) by combining retrieved task-relevant tutorials and look-ah
Large language models (LLMs) can memorize and reproduce fragments of their training data, raising privacy and
Reinforcement learning with verifiable rewards (RLVR) can enable generalizable improvement for large language
Identifying ligands for a protein target requires searching a vast chemical space for a comparatively small se
**Purpose:** Ultrasound biometry is widely used across various clinical applications, including prenatal and c
Zeroth-order (ZO) optimization enables memory-efficient large language model (LLM) fine-tuning, and recent met
Zeroth-order (ZO) optimization has emerged as a promising approach for memory-efficient large language model (
Previous research shows that vision-language (VL) models struggle to understand fine-grained image details and
Dhārā Bench is a retrieval benchmark for Indian Knowledge Systems (IKS), with 325 questions in 12 categories a
Large language agents for literary role-playing are typically modeled around dialogue generation, leaving char
Scaling depth is a key route to stronger large language models, but Pre-LN architectures often suffer from dep
We propose a simple, linguistically motivated approach to detecting AI-generated peer reviews based on the inh
This paper studies retrieval-augmented generation (RAG), which leverages external knowledge to improve the per
Large language models are increasingly consulted for religious knowledge, including the recall of specific Had
Automated liver ultrasound report generation improves the diagnosis of liver diseases by producing clinically
Retrieval-augmented generation (RAG) extends large language models with external evidence, yet relevant eviden
Generating visualisations from natural language requests (text-to-vis) enables users to explore data more effi
AI assistants store values that a language model derived from a person's records under the person's own defini
Humans imagine scene changes under hypothetical motions, an ability central to spatial reasoning and robotic p
Text-to-CAD generation aims to make CAD modeling more accessible by generating executable CAD programs from na
Source-only domain generalization (DG) for remote sensing semantic segmentation seeks reliable predictions in
Geospatial text spotting involves locating and transcribing geographic labels, such as road names, district na
Multimodal large language models (MLLMs) have shown strong potential in robotic surgery for question answering
Multi-source remote sensing data classification has received growing attention recently. However, existing met
Cloth-changing person re-identification (CC-ReID) is still very challenging, and no method clearly dominates:
Widespread image sharing on social media has increased the need for visual privacy recognition to identify ina
Detecting defects in real-world industrial inspection often requires learning from only a few normal images pe
Cross-view object correspondence (CVOC) aims to establish associations of the same target between query and se
Progress in robot learning, particularly physical multi-agent reinforcement learning (MARL), is currently chal
On-policy distillation (OPD) can improve answer accuracy while weakening responsiveness to supporting evidence
Query-conditioned sparse KV selection reduces long-context inference cost by attending to a subset of historic
Web communities rely on moderators to catch harmful messages such as insults, hate speech, and scams, but no o
Electroencephalography (EEG) provides a non-invasive means for neurological and mental health assessment, yet
Vision-language models for remote sensing answer questions about satellite images by generating text: every an
Post-trained language models fabricate citations at high rates. We ask whether that is reward hacking: scoring
Vision-language models can interpret instructions and plan manipulation tasks, yet struggle when acting direct
Concept erasure is a prominent approach to achieving fairness in machine learning through the removal of sensi
Camera localization methods are mostly multi-stage pipelines relying on dedicated and complex maps that are co
Time-series forecasters are ranked almost entirely by accuracy metrics, but a forecast's decision value, its e
Prediction models degrade when the deployment population differs from the one used to train them, and training
Phone scams impose substantial and growing costs on individuals and financial systems worldwide, yet existing
Black-box optimization (BBO) arises in many scientific and engineering problems where objective evaluations ar
Frameworks that use large language models for scientific discovery typically rely on a fixed, human-designed a
Spiking neural networks (SNNs) offer low-energy sequence modeling through sparse, event-driven computation. Ho