The paper evaluates a multilingual ASR model (MMS‑1B‑all) on a Garrusi Kurdish dataset using a common‑reference staged normalization approach. By normalizing both reference and hypothesis, the authors show that raw Arabic‑script hypotheses yield a 111.70 % WER, which drops to 97.85 % after folding into a reduced orthography, highlighting the impact of orthographic differences on error measurement. A Southern Kurdish fine‑tuned system performs worse, and residual errors are partly due to scoring‑pipeline limitations rather than recognition failures.
By Hiwa Asadpour
SenseNova-U1.5 is an 8B‑MoT native unified multimodal model that can understand, reason about, and generate visual content without using an encoder or VAE. It improves visual fidelity and text rendering through spatially coherent patch reconstruction, large‑scale training on curated generation and editing data, and native resolutions up to 4K. Post‑training, specialized experts for visual aesthetics, bilingual text rendering, infographic generation, and image editing are optimized and distilled into a multi‑expert framework, yielding advances in image fidelity, complex composition, multi‑reference editing, and instruction following.
By Haiwen Diao, Jiahao Wang, Chenjing Ding, Hanming Deng, Jiangnan Chen, Ruixi Zhang, Ruohui Wang, Wenwen Tong, Xiangyu Fan, Yubo Wang, Yue Zhu, Yuwei Niu, Zhengqi Bai, Zhiqian Lin, Zhitao Yang, Zhongang Cai, Bo Yang, Chen Feng, Chengguang Lv, Guangjia Liu, Guanlin Wang, Hanyu Zhang, Haojia Yu, Hongcan Xiao, Hongli Wang, Huan Wu, Huaping Zhong, Jian Fang, Jianan Fan, Jiaqi Li, Jiefan Lu, Jing Zuo, Jingcheng Ni, Junxiang Xu, Linjun Dai, Mutian Xu, Peishen Yan, Penghao Wu, Ruijie Mao, Ruisi Wang, Shihao Bai, Shuang Yang, Shuya Yang, Shuyan Zheng, Silei Wu, Siying Li, Tao Chu, Tianbo Zhong, Tongxi Zhou, Weichao Luo, Weichen Fan, Wenhao Jia, Wenjie Gao, Xiangli Kong, Yan Li, Yang Yong, Zimo Wen, Zixuan Qian, Wenxiu Sun, Ruihao Gong, Quan Wang, Lewei Lu, Lei Yang, Ziwei Liu, Dahua Lin
The paper introduces the task alignment proxy, a method that accelerates hyperparameter selection for merging models fine‑tuned on diverse vision tasks. It addresses the challenge of training heterogeneous decoders, which makes traditional downstream performance evaluation costly. By using the proxy, the authors demonstrate that model merging can be applied efficiently to multi‑task vision models beyond CLIP‑based classification.
By Pau de Jorge, C\'esar Roberto de Souza, Bj\"orn Michele, Mert B\"ulent Sar{\i}y{\i}ld{\i}z, Philippe Weinzaepfel, Florent Perronnin, Diane Larlus, Yannis Kalantidis
SPECTRA is a parameter‑efficient fine‑tuning framework for geospatial foundation models that tackles two key challenges: spectral mismatch and high adaptation cost. It introduces Band‑Routed Embedding (BRE) to map downstream sensor bands into the pretrained model’s expected band space, enabling full use of available spectral data without altering the patch embedding interface. Additionally, Stage‑wise Transferability‑aware LoRA (ST‑LoRA) estimates stage‑wise transferability and assigns LoRA ranks accordingly, concentrating trainable parameters on the most transferable stages and reducing overall adaptation cost.
By Xingyan Li, Jordan A. Caraballo-Vega, Jie Gong, Mark L. Carroll, Jianwu Wang
The study compares two ways of obtaining predictions from language models fine‑tuned on customer behavior: scoring answer tokens directly versus generating a written rationale and then scoring the resulting answer. Across 13 model‑domain cells covering four retail tasks, scored readouts consistently rank outcomes more accurately than generated readouts, with an AUC improvement ranging from 1.5 to 14.5 points. The authors also find that a third readout—eliciting a probability before any verdict—improves calibration but only when outcome rates are represented in training, and they recommend using generated rationales for interpretability while relying on scored heads for ranking.
By Touchapon Kraisingkorn, Krittin Pachtrachai, Wachiravit Modecrua
LLMAR is a tuning‑free recommendation framework designed for sparse, text‑rich industrial B2B domains. It transforms user behavioral history into structured semantic motives using LLM inference, employs a reflection loop to self‑correct hallucinations, and operates cost‑effectively with asynchronous batch processing. Experiments on MovieLens‑1M, Amazon Prime Pantry, and a construction risk dataset show LLMAR surpasses state‑of‑the‑art learning models, achieving up to a 54.6% nDCG@10 improvement while keeping inference costs around $1 per 1,000 users.
By Ryogo Hishikawa, Ichiro Kataoka, Shinya Yuda
The paper introduces a method that learns dynamic reaction representations directly from textual descriptions using a fine‑tuned language model coupled with Gaussian process surrogates. This approach enables multi‑objective Bayesian optimisation for chemical reactions, achieving faster convergence than traditional descriptor libraries or one‑hot encodings across nickel‑, palladium‑, and iridium‑catalysed systems. Prospective experiments on a palladium‑catalysed cyanation and an asymmetric hydrogenation produced high‑yield, high‑enantiomeric‑excess conditions after only two rounds of high‑throughput testing, translating directly to gram‑scale synthesis.
By Joshua W. Sin, David Ming Segura, Bojana Rankovi\'c, Siu Lun Chau, Marius D. R. Lutz, Andrea Anelli, Ryan P. Burwood, Kurt P\"untener, Maximilian J. Notheis, Raphael Bigler, Philippe Schwaller
The paper introduces Prototype-Mediated Process Supervision (PMPS), a method that uses learnable reasoning prototypes to provide structural supervision for latent chain-of-thought embeddings, addressing representation collapse and uneven information distribution. PMPS aligns latent and explicit CoT embeddings in a shared prototype space via many-to-many soft assignment and employs a Progressive Sequential Alignment module to guide training from positional priors to adaptive matching. Experiments show PMPS reduces output token length to under 50% of explicit CoT on GSM8K-Aug and improves accuracy by 2.08% over SIM-CoT, even surpassing CoT-SFT on GPT-2 and achieving the highest accuracy among latent reasoning methods on larger models and harder tasks.
By Yiqi Li, Xu Chen, Chen Ju, Jiangchao Yao, Zhaoyang Li, Jinsong Lan, Xiaoyong Zhu, Bo Zheng, Yu Wang
The study investigates how fine‑tuning large language models on synthetic stories can imprint human character traits onto AI assistants. Even when only a small fraction of stories contain a particular behavior, the assistant adopts that conditional behavior while remaining generally helpful. The researchers find that the assistant is more influenced by characters that resemble its own persona—an effect they call the affinity effect—and that this influence extends to base models and different system prompts.
By Jorio Cocola, Lev McKinney, Harry Mayne, Jan Betley, Owain Evans
The paper introduces Structurally Speaking, a lightweight structured prompting protocol that improves graph captioning by translating explicit connectivity into concise motif-level descriptions. It treats motif-oriented graph captioning as a bidirectional graph-text translation task, aiming to preserve topology for graph recovery while producing interpretable captions. Experiments on a synthetic motif-based dataset show that the protocol yields shorter, more motif-consistent captions without sacrificing graph recovery performance.
By Hsiao-Ying Lu, Dongyu Liu, Kwan-Liu Ma
The paper proposes an information-weighted cross‑entropy loss that rescales token contributions using TF‑IDF statistics, thereby emphasizing semantically informative tokens and down‑weighting ubiquitous ones. Experiments on five decoder‑only language models (1.1B–13B parameters) show consistent reductions in memorized substring length while maintaining perplexity and downstream performance. The method is architecture‑agnostic, adds less than 3% computational overhead, and can be integrated into existing training pipelines.
By Zhijian Li, Stefan Larson, Kevin Leach
LILA (Latent-Informed Layer Analysis) introduces a calibration‑free method for structured pruning of large language models by scoring neuron importance using the Kolmogorov–Smirnov distance between singular value distributions of full and neuron‑ablated feed‑forward network weight matrices. The approach requires no training, calibration data, or auxiliary networks, and outperforms existing methods such as PruneNet and SliceGPT on LLaMA‑2‑7B and Phi‑2 at various sparsity levels. After a single epoch of LoRA fine‑tuning, LILA matches heavily calibrated baselines, and a Neural Tangent Kernel analysis provides theoretical support for its spectral importance criterion. Additionally, LILA can dynamically allocate sparsity budgets, achieving state‑of‑the‑art generative preservation and revealing architectural bottlenecks at higher compression.
By Sankar Behera, Dhruv Singh, Anshika Agnihotri, Raj Kumar Choudhary, Satyadev Ahlawat, Yamuna Prasad
CMNIE is a new benchmark for extracting structured information from Chinese military news, covering event triggers, arguments, named entities, and entity relations under a unified schema. The dataset contains 13,000 manually annotated instances with 7 event types, 10 argument roles, 7 entity types, and 8 relation types. Experiments show that current supervised models, zero‑shot LLMs, and fine‑tuned LLMs struggle with relation extraction and exact span matching, highlighting the challenge of joint structured extraction in this domain.
By Yan Yu, Mengna Zhu, Zhenyu Song, Hao Yang, Haiwen Chen, Mao Wang
The paper introduces a new automatic evaluation metric for human simultaneous interpreting that aligns with traditional analytic rubrics. A corpus of 1,101 professionally scored SI segments is created, covering meaning transfer, delivery quality, and perceived latency. Using a LoRA‑adapted COMET‑KIWI encoder with dual regression heads, the model achieves Pearson correlations of 0.388 for meaning transfer and 0.301 for delivery quality, outperforming the frozen baseline while acknowledging low rater agreement.
By Ziyu Zhang, Satoshi Nakamura
The paper presents a pipeline that leverages large language models to extract grammatical rules, example sentences, and lexicons from descriptive grammar books, producing synthetic parallel corpora for fine‑tuning machine translation models. Evaluated on three low‑resource languages—Kalamang, Tuatschin, and Mandan—the synthetic data improves translation quality over seed‑data baselines in 75% of configurations for Kalamang and 59% for Tuatschin, achieving up to +8.8 ChrF++ gains. A factorial study across 96 configurations identifies which combinations of target part‑of‑speech, retrieval granularity, and sample volume drive performance gains and where they fail, demonstrating that static linguistic documentation can be repurposed for practical translation tools for severely under‑resourced languages.
By Varun Ghat Ravikumar, Sina Ahmadi, Lena J\"ager, Rico Sennrich
The paper introduces SHAPER, a self‑evolving framework that enables embodied agents to adapt to new environments without retraining the underlying foundation model. SHAPER keeps model parameters frozen and instead evolves reusable skills and a context‑code harness through rollouts in the target environment, allowing the same model to act as both planner and optimizer. Experiments on VLABench and ESI‑Bench demonstrate that this skill‑and‑harness optimization outperforms pure execution, supervised fine‑tuning, and test‑time scaling baselines, showing that self‑evolving adaptation is viable when model training is costly or impractical.
By Peidong Wang, Zhiming Ma, Ying Chang, Xufang Luo, Yiqun Zhang, Zihan Wang, Xiaocui Yang, Shi Feng, Yuqing Yang, Dongsheng Li
The paper introduces two lightweight spectral adapters—Directional Spectral Adapter (DiSECT) and Spectral Instance-Guided Adapter (SiGA)—to adapt the Segment Anything Model (SAM) for accurate segmentation of colorectal liver metastases in contrast‑enhanced CT scans. SiGA achieves the highest single‑point Dice score of 0.77 and performs comparably to a 3D nnU‑Net baseline under no‑prompt inference, while DiSECT requires only 0.14 million trainable parameters. The study evaluates the adapters on 446 CT volumes across various prompting regimes, demonstrating that spectral adapters can efficiently adapt SAM with limited trainable parameters while maintaining strong segmentation accuracy.
By Ramtin Mojtahedi, Mohammad Hamghalam, Jacob J. Peoples, Natalie Gangai, Mithat Gonen, Yun Shin Chun, HyunSeon Christine Kang, Richard K. G. Do, Amber L. Simpson
HALDETECT is a system developed for the English hallucination-detection track of ImageEval 2026, where the task is to identify the single visually grounded statement among three culturally plausible options. The approach treats the problem as a contrastive decision, outputs the answer before an explanation, and bases reasoning on colour/texture, shape/form, and context. The best model fine‑tunes Qwen2.5‑VL‑7B‑Instruct with 4‑bit QLoRA, freezes the vision encoder, and achieves a Contrastive Instability score of 0.035 on the test set, placing third among eight teams.
By Syed Mohaiminul Hoque, Md Sakhawat Hossain
The paper introduces LASP, a framework that uses large language models to generate structured B‑spline priors from procedural modeling histories. By translating design intent into rich textual descriptions, LASP provides semantic reasoning that guides conventional CAD solvers toward more accurate and coherent surface fitting. Experiments show that language‑driven priors outperform traditional machine learning approaches, establishing a new paradigm for language‑guided geometric optimization.
By Yunzhong Lou, Yusheng Luo, Jiahao Li, Yu Song, Xiangdong Zhou
The paper introduces a multi‑signal pipeline for detecting hallucinations in large language model outputs, combining fine‑tuned DeBERTa‑v3 classification, Monte Carlo Dropout uncertainty, and temperature‑scaled calibration. On the HaluEval benchmark it achieves strong performance (F1 = 0.915, AUROC = 0.977) and further improves accuracy to 93.2% with MC Dropout. The authors also demonstrate that applying Direct Preference Optimization to a Qwen2.5‑0.5B generator reduces hallucination rates from 85.5% to 37.7%, and show that domain‑specific fine‑tuning (PubMedBERT on SciFact) yields better results than general‑domain training.