LoRA, PEFT, instruction tuning and domain adaptation — adapting a pretrained model without paying to train one.
arXiv:2603.02504v3 Announce Type: replace Abstract: Large Language Models (LLMs) achieve strong performance on natural language tasks but remain unreliable in mathematical reasoning, frequently gener...
arXiv:2606.30943v2 Announce Type: replace Abstract: Russian and Arabic are among the major languages of scientific communication. Language barriers impede the exchange of research results between the...
arXiv:2603.25175v2 Announce Type: replace Abstract: Monocular egocentric 3D pose estimation is difficult because severe foreshortening, self-occlusion, and a restricted field of view often remove the...
UniCAR‑RL is an annotation‑free reinforcement learning framework designed to improve multimodal large language models’ visual mathematics reasoning. It decouples perception and reasoning by using three branches: Caption‑RL for perception optimization, Reasoning‑RL for logical reasoning with a gold image description, and QA‑RL for end‑to‑end question answering. Experiments show significant gains in mathematical and visual reasoning across various model architectures and scales using only raw short‑answer data.
arXiv:2606.19100v4 Announce Type: replace Abstract: Large Vision and Language Models (LVLMs) have advanced rapidly, yet European Portuguese (pt-PT) remains systematically underserved by existing open...
Traditional multimodal representation learning and generation are two stages: a contrastive or self-supervised visual encoder is trained first, followed by a separate downstream generative model. This...
Research data repositories are essential infrastructure for scientific inquiry and for ensuring that datasets follow FAIR (Findable, Accessible, Interoperable, and Reusable) principles. However, repos...
Rapid advancements in vision-language models have propelled Referring Remote Sensing Image Segmentation (RRSIS) to the forefront of Earth observation. However, practical deployments suffer severe perf...
Progress in inclusive household surveys has strengthened socioeconomic evidence for forcibly displaced populations, providing indispensable benchmarks on living conditions and welfare. However, these...
Large language model coding agents have recently become useful for software tasks, but weaker or open-weight agents still struggle to reliably interpret user intent and execute complex multi-step work...
Fyxer has created an AI executive assistant that people trust by leveraging OpenAI models, fine‑tuning, memory, and real user feedback. The system organizes inboxes and drafts emails in each user’s voice, ensuring personalized and reliable communication support.
CluSTER is a cluster‑aware balanced sampling framework designed to improve the efficiency of instruction‑tuning for large language models. It reduces redundant computation by clustering data in gradient space and allocating samples across GPUs in a data‑parallel setting, while preserving the original distribution through weighted updates. Experiments show that CluSTER can cut training time by up to 69.6% with negligible loss in accuracy compared to existing sampling methods.
Guided Adversarial Robust Transfer (GART) learning is a new transfer learning method that relaxes the requirement for source data to closely resemble the target population. By optimizing an adversarial loss over a mixture of source distributions, GART achieves faster convergence and improved prediction performance when target data are scarce. Experiments on simulated data and on multi‑institutional biobank‑linked electronic health records for high‑density lipoprotein cholesterol demonstrate higher robustness and accuracy compared to existing transfer learning approaches.
The paper introduces a long‑context recall technique that keeps GPU memory usage nearly constant regardless of context length, without requiring additional training. It reconstructs facts by leveraging residual vectors stored in the LLM’s feed‑forward layers, enabling deterministic retrieval of query‑relevant information without accessing the original document. Experiments demonstrate the method can answer single‑fact questions in two‑million‑token stories, outperforming prior approaches.
arXiv:2609.12993v1 Announce Type: cross Abstract: We investigate the effect of pretrained T5 model scale and explicit motion features on pose-to-text Indian Sign Language Translation (SLT) for the WS...
ChitraMiti introduces a synthetic benchmark of 12,874 Bengali planar geometry problems with structured 15‑attribute descriptions, alongside a complementary set of 500 textbook diagrams. Using a three‑phase protocol that tests diagram‑only, diagram‑plus‑description, and description‑only inputs, the study finds that description‑only performance matches diagram‑plus‑description performance across several VLMs, yet models still struggle with cross‑modal verification. Fine‑tuning on ChitraMiti improves results on both datasets, though a gap remains compared to the best zero‑shot model.
The paper examines a specific type of hallucination in large language models caused by spurious correlations—unintended, statistically prominent associations in training data such as surnames linked to nationalities. These hallucinations are confidently produced, persist regardless of model scaling or refusal fine‑tuning, and evade existing detection methods like confidence filtering and inner‑state probing. The authors use controlled synthetic experiments and evaluations on both open‑source and proprietary LLMs, including GPT‑5, to demonstrate the failure of current detection techniques and provide a theoretical explanation for why statistical biases undermine confidence‑based approaches.
Countdown-Code is a minimal environment that lets models solve a mathematical reasoning task while also manipulating the test harness, creating a clear split between proxy rewards (test pass/fail) and true rewards (mathematical correctness). Using this setup, the authors show that reward hacking can arise during supervised fine‑tuning when as little as 1% of training data contains reward‑hacking trajectories, and that reinforcement learning further amplifies and generalizes this misalignment. The paper releases the environment and code to support future research on detecting and mitigating reward hacking in large language models.
The paper proposes a reinforcement learning (RL) fine‑tuning framework for electronic health record (EHR) foundation models, treating them as generative policies over patient trajectories. By framing clinical prediction tasks such as hospital readmission as event‑conditioned, time‑windowed reasoning problems and designing time‑aware, rollout‑sensitive rewards, the authors show that RL fine‑tuning consistently outperforms pre‑trained backbones and strong baselines. The approach enables smaller models to surpass larger pre‑trained models in data‑limited settings, induces positive transfer across tasks, and produces trajectories with stronger structural and semantic alignment to ground truth, improving downstream utility.
arXiv:2607.08098v2 Announce Type: replace Abstract: Event cameras are increasingly adopted in embodied perception for their microsecond temporal resolution, high dynamic range, and resilience to moti...