Tailor3D leverages dual-side images and LoRA-based triplane Transformer for fast, personalized 3D asset editing, achieving high consistency and efficiency.

Zhangyang Qi, Yunhan Yang, Mengchen Zhang et al.

Proposes TARGO-Net with shape completion and Transformer architecture, achieving robust target grasping under occlusion, evaluated on large synthetic and real datasets.

Yan Xia, Ran Ding, Ziyuan Qin et al.

Proposes MDT, a diffusion transformer that learns versatile multimodal goal behaviors with minimal language annotations, achieving 15% improvement on CALVIN benchmarks.

Moritz Reuss, Ömer Erdinç Yağmurlu, Fabian Wenzel et al.

LLaMAX leverages continual multilingual pretraining and vocabulary preservation to support over 100 languages, achieving +10 spBLEU over open-source models, comparable to M2M-100-12B.

Yinquan Lu, Wenhao Zhu, Lei Li et al.

Proposes TSGi, integrating scene graphs with CLIP-based multimodal learning, achieving 57.77% balanced accuracy in traffic accident classification, a near 5% improvement.

Aaron Lohner, Francesco Compagno, Jonathan Francis et al.

GenArtist employs a multimodal large language model as an agent for unified image generation and editing, surpassing DALL-E 3 with over 7% improvements on benchmarks.

Zhenyu Wang, Aoxue Li, Zhenguo Li et al.

Permutation Equivariant neural optimizer LOOP-PE enhances multi-sensor decision-making, achieving 19x faster solutions in real-time energy management.

Meiyi Li, Javad Mohammadi

This study shows that advanced language models, via linear mapping, outperform traditional ID-based recommenders, revealing a space isomorphism and implicit collaborative signals.

Leheng Sheng, An Zhang, Yi Zhang et al.

Using LTL to generate 2000 challenges, evaluating 12 LLMs, revealing three main issues in complex temporal reasoning tasks.

Weizhi Tang, Kwabena Nuamah, Vaishak Belle

Emilia is a large-scale, multilingual speech dataset (>101k hours) with an open-source pipeline, Emilia-Pipe, enabling efficient processing of wild speech for natural speech generation.

Haorui He, Zengqiang Shang, Chaoren Wang et al.

Test-time contrastive concepts improve open-vocabulary semantic segmentation by automatically generating query-specific negative samples, boosting accuracy.

Monika Wysoczańska, Antonin Vobecky, Amaia Cardiel et al.

Constructed MMSci dataset with 72 disciplines, enabling large models to understand complex scientific figures; fine-tuned Qwen2-VL-7B achieved 87.48% accuracy.

Zekun Li, Xianjun Yang, Kyuri Choi et al.

This paper introduces debate as a scalable oversight protocol, using weak LLMs to judge strong LLMs, showing superior performance in information-asymmetry tasks with up to 8% accuracy gain.

Zachary Kenton, Noah Y. Siegel, János Kramár et al.

OneRestore framework uses cross-attention to restore complex image degradations, improving PSNR and SSIM.

Yu Guo, Yuan Gao, Yuxu Lu et al.

AriGraph significantly enhances LLM agents' task handling by integrating semantic and episodic memory.

Petr Anokhin, Nikita Semenov, Artyom Sorokin et al.

A comprehensive review of LLM evaluation challenges, emphasizing reproducibility, reliability, and robustness, with specific algorithm and dataset references.

Md Tahmid Rahman Laskar, Sawsan Alqahtani, M Saiful Bari et al.

ComplexBench benchmark with hierarchical taxonomy and rule-augmented evaluation reveals significant deficiencies of current LLMs in multi-constraint complex instructions.

Bosi Wen, Pei Ke, Xiaotao Gu et al.

Proposed a classification-based framework with multi-round search, boosting HDL code correctness by over 14% in Pass@10 on VerilogEval.

Wenhao Sun, Bing Li, Grace Li Zhang et al.

NEBULA combines VQ-VAE and neural empirical Bayes for rapid, scalable 3D molecular library generation with strong generalization.

Ewa M. Nowara, Pedro O. Pinheiro, Sai Pooja Mahajan et al.

AMEX dataset enhances mobile GUI agent performance with 104K screenshots and multi-level annotations.

Yuxiang Chai, Siyuan Huang, Yazhe Niu et al.