LoRA, PEFT, instruction tuning and domain adaptation — adapting a pretrained model without paying to train one.
arXiv:2609.26388v1 Announce Type: cross Abstract: Recent advances in large language models (LLMs) have made them widely used for code-related tasks. Identifier names are statistically informative in...
arXiv:2601.20900v4 Announce Type: replace-cross Abstract: Adapting large language model (LLM)-based automatic speech recognition (ASR) systems to new domains using text-only data is a significant yet...
arXiv:2609.26536v1 Announce Type: new Abstract: In LLM-based speech translation, transcription-based chain-of-thought (CoT) suffers from a mismatch between reference transcripts used in supervised fi...
arXiv:2609.25176v1 Announce Type: cross Abstract: Real-time voice assistants must reason over evolving requests, execute actions, and follow conversational rules. Qwen-Audio-3.1-Realtime brings these...
arXiv:2608.24268v2 Announce Type: replace Abstract: This paper proposes methods to extract over 50 types of events from a Dutch historical corpus spanning the 17th and 18th centuries. The methods we...
arXiv:2609.26189v1 Announce Type: new Abstract: Retinal vessel segmentation in multi-domain deployment requires a source model to adapt to domains that differ in imaging conditions and annotation con...
arXiv:2609.26443v1 Announce Type: new Abstract: Recent advances in Artificial Intelligence (AI)-powered Computer-Aided Diagnosis (CAD) systems have substantially improved breast cancer screening, dia...
arXiv:2609.25575v1 Announce Type: cross Abstract: Fine-tuned Large Language Models (LLMs) significantly advance Automated Theorem Proving (ATP), but are often deployed as guiding policies within tree...
EDGEGEN is a synthetic task generation framework that extracts compliance rules from a tool‑calling agent’s specification to create database‑grounded edge‑case tasks that violate those rules. By combining EdgeGen with existing synthetic data generation methods, it forms a fully automated closed‑loop system that requires no human annotation. Experiments show that finetuning on EdgeGen data improves performance by 2–42 % on the tau2bench airline domain, while harness optimization yields 10–30 % gains over human‑curated and base harnesses for the Gemma‑4‑e4b model.
WILSON is a vision–language foundation model that represents whole‑slide images and multi‑slide patient cases as single multi‑magnification composite images. Trained on about 189,000 Mayo Clinic slides covering 42 organs and 829 diagnostic entities, it outperforms dedicated case‑level models on internal cohorts and matches slide‑level models while using far less compute. Fine‑tuning on triple‑negative breast cancer data improves histologic subtyping and lymphocyte grading, and the model retrieves diagnostic text with high recall and generates captions closer to report references than prior methods.
The ICDAR2026 Competition on Multimodal Reasoning over Documents in Multiple Domains introduced a new Visual Question Answering benchmark that tests reasoning over documents from eight distinct domains such as business reports, scientific papers, and engineering drawings. Twenty valid submissions from eight teams were evaluated, featuring approaches ranging from zero‑shot vision‑language models to multi‑agent ensembles and fine‑tuned multimodal systems. Results indicate that the most effective systems employ structured evidence extraction, retrieval, verification, and orchestration across multiple components rather than single‑pass prompting.
The paper introduces modality‑gated deep adapters, a parameter‑efficient method for adding new modalities to a frozen multimodal embedding language model without altering its existing outputs. These adapters are bottleneck modules attached to each decoder layer, grouped into modality‑specific packs that activate only during encoding of their own modality, ensuring exact preservation of the base model’s computation graph. Experiments on a 2B base model show significant gains in audio‑to‑text and thermal‑to‑text retrieval metrics, and the authors release the audio and thermal packs along with training and evaluation code.
The paper introduces a framework for estimating peer effects using multivariate behavioral measures extracted from written text via a large language model (LLM). It demonstrates that LLM embeddings enhance out‑of‑sample recidivism prediction by up to 30% compared to pre‑entry covariates alone, and presents a novel instrumental variable estimator that is √N‑consistent for sparse networks with multidimensional latent homophily. By combining limited human annotations with LLM zero‑shot vectors, the authors develop a prediction‑powered peer inference method that yields de‑biased estimates and reveals significant peer effects in behavioral profiles.
LIMIT (Less Is More for Instruction Tuning in Text-to-SQL) challenges the belief that large instruction corpora are necessary for effective Text-to-SQL models. The framework uses a four‑stage data‑centric process—difficulty‑aware filtering, chain‑of‑thought synthesis, LLM‑as‑judge quality scoring, and genetic algorithm optimization—to select a compact set of examples that still achieve full schema coverage. On the BIRD and Spider benchmarks, LIMIT’s 796 and 863 samples enable Qwen3‑8B to reach 69.1% and 88.9% execution accuracy, outperforming methods trained on twenty times more data and setting a new state‑of‑the‑art for open‑source approaches.
The paper introduces WaterBERT, a domain‑adapted encoder model trained on a 2.97‑billion‑token water treatment corpus to capture domain‑specific semantics for literature mining. Fine‑tuned versions of WaterBERT outperform general‑purpose and other domain BERT models on tasks such as treatment process classification, named entity recognition, and relation extraction. The authors also demonstrate WaterBERT’s utility in large‑scale processing, generating coherent research topics, building a structured knowledge graph from 693,211 abstracts, and creating a Water Knowledge‑Enhanced Retrieval System that surpasses text‑based baselines.
CONCAT is a training‑free framework that improves the efficiency of large language model (LLM) based multi‑agent systems by clustering agents according to their initial answers and selecting cluster leaders based on confidence. It uses a Theory‑of‑Mind‑inspired heuristic to predict collaboration benefits between leaders, then prunes communications to form an ad‑hoc network that reduces latency. Experiments on three LLMs and benchmarks show up to 2.02× higher accuracy/latency ratio than LLM‑Debate and a 50.1% latency reduction on Qwen2.5‑14B‑Instruct without task‑specific training.
The paper introduces a ptychography‑supervised learning framework that transforms 4D‑STEM data into real‑time atomic‑resolution phase images. By training a compact model on physics‑constrained reference phase maps from a single AuPd dataset, the method predicts local phase patches directly from diffraction patterns without probe calibration or iterative optimization. The resulting workflow achieves an online latency of ~0.27 ms per probe position, a 1,000‑fold speed‑up over GPU‑accelerated ePIE, and maintains atomic‑scale lattice contrast while generalizing across materials, defocus conditions, and instruments.
Metric-Bench introduces a new benchmark for Vision‑Language Models (VLMs) that focuses on metric‑spatial reasoning in indoor scenes by using in‑image reference objects with known dimensions. The accompanying MetricReasoner fine‑tuning recipe employs structured prompts and numerical rewards to implicitly learn 2D‑to‑3D mapping without camera intrinsics. Experiments show that this approach improves spatial metric understanding by 43.1 % over existing models and boosts downstream embodied tasks, while also delivering gains on general VLM benchmarks.
arXiv:2609.22529v1 Announce Type: new Abstract: International law provides the normative framework through which states coordinate action, regulate armed conflict, and protect human rights, yet its t...
arXiv:2606.28798v2 Announce Type: replace Abstract: The main objective of this paper is to propose a general framework for prediction based on different sources of multimodal data in the healthcare d...