Explore · 每周精选
发现最受关注的研究论文,追踪研究趋势,订阅感兴趣的期刊与关键词。
598 篇论文 · 按点赞排序
Xianjie Wu, Jian Yang, Linzheng Chai +10 authors
TableLLM, trained on TableInstruct, benchmarks performance in tabular data reasoning and highlights the need for improvement in real-world applications.
Zekun Wang, King Zhu, Chunpu Xu +14 authors
MIO, a novel foundation model, achieves competitive performance in multimodal tasks through an end-to-end, autoregressive approach using causal multimodal modeling.
Pratyush Maini, Skyler Seto, He Bai +3 authors
Web Rephrase Augmented Pre-training (WRAP) improves large language model pre-training efficiency and performance by incorporating synthetic rephrases styled like Wikipedia or question-answer format into the training dataset.
Yuqing Wang, Shuhuai Ren, Zhijie Lin +6 authors
A parallel generation strategy for autoregressive models improves inference speed without significantly compromising quality in visual generation tasks.
LG AI Research, Soyoung An, Kyunghoon Bae +30 authors
EXAONE 3.5 language models, available in three configurations, demonstrate exceptional instruction following, long-context comprehension, and competitive performance across various benchmarks.
Duy-Kien Nguyen, Mahmoud Assran, Unnat Jain +3 authors
Vanilla Transformers achieve high performance in computer vision tasks by treating individual pixels as tokens, challenging the conventional use of local neighborhoods.
Jinbin Bai, Tian Ye, Wei Chow +6 authors
Meissonic, a non-autoregressive masked image model, achieves high-quality image generation by incorporating advanced techniques and outperforms existing models like SDXL.
Fuzhao Xue, Yukang Chen, Dacheng Li +15 authors
LongVILA, a full-stack solution for long-context vision-language models, introduces Multi-Modal Sequence Parallelism for efficient training and inference, and a five-stage training pipeline that enhances long video processing and captioning performance.
Haoran Li, Qingxiu Dong, Zhengyang Tang +17 authors
GLAN, a general method for instruction tuning of LLMs, uses a taxonomy of human knowledge to generate synthetic instruction data, achieving strong performance across diverse tasks without task-specific training data.
Xiaoding Lu, Adian Liusie, Vyas Raina +2 authors
A combination of smaller chat AI models, through a method called blending, can achieve performance comparable to or better than large single models with lower computational resources.
Clément Christophe, Praveen K Kanithi, Tathagata Raha +2 authors
Med42-v2 enhances Llama3 with clinical data to improve performance in healthcare settings, outperforming generic models across medical benchmarks.
Shaofei Cai, Zihao Wang, Kewei Lian +4 authors
Visual-temporal context prompting integrates object segmentation and visual observations to enhance VLMs in embodied decision-making, enabling complex spatial tasks.
Jiayuan Zhu, Yunli Qi, Junde Wu
MedSAM-2 extends the SAM 2 framework for 2D and 3D medical image segmentation, offering one-prompt segmentation across multiple images and outperforming existing models in various medical imaging tasks.
Adam Ibrahim, Benjamin Thérien, Kshitij Gupta +5 authors
Simple continual learning strategies, including learning rate re-warming, re-decaying, and data replay, can match the performance of full re-training for large language models with much lower computational cost.
Tianbao Xie, Danyang Zhang, Jixuan Chen +14 authors
OSWorld provides a scalable, real computer environment for evaluating multimodal agents across various operating systems and tasks, highlighting significant gaps in current LLM/VLM capabilities.
Yuri Kuratov, Aydar Bulatov, Petr Anokhin +4 authors
BABILong benchmark evaluates large language models' reasoning across long documents, revealing that models use only a small portion of context and that recurrent memory transformers handle longer contexts better.
Eldar Kurtic, Alexandre Marques, Shubhra Pandit +2 authors
A comprehensive study evaluates quantization formats for LLMs, showing that FP8 W8A8-FP is lossless, W8A8-INT incurs minimal accuracy loss, and W4A16-INT is cost-efficient, providing guidelines for optimal deployment.
Xuekai Zhu, Daixuan Cheng, Hengli Li +7 authors
The use of synthetic data in language model training leads to model collapse, which is mitigated by token-level editing of human-produced data to create semi-synthetic data.
Liang Zeng, Liangjun Zhong, Liang Zhao +9 authors
Skywork-Math model series achieves high performance in math reasoning by fine-tuning on a large and diverse dataset, demonstrating that data quantity significantly enhances LLM mathematical capabilities.
Gemma Team, Thomas Mesnard, Cassidy Hardin +105 authors
Gemma, a family of lightweight and high-performing language models, outperforms similarly sized open models across text-based tasks and emphasizes the importance of responsible model development and safety.
Arindam Mitra, Luciano Del Corro, Guoqing Zheng +11 authors
AgentInstruct, an agentic framework, generates high-quality synthetic data for post-training language models, leading to significant performance improvements across various benchmarks.
Yu Lu, Yuanzhi Liang, Linchao Zhu +1 authors
FreeLong enhances long video generation by balancing global and local video features during denoising, improving quality and consistency.
Xin Huang, Tengfei Wang, Ziwei Liu +1 authors
Material Anything is a unified diffusion framework that generates physically-based materials for 3D objects using a pre-trained image diffusion model with enhancements for stability and quality.
Manuel Faysse, Hugues Sibille, Tony Wu +3 authors
The ColPali retrieval model uses Vision Language Models to create embeddings from images of document pages, improving performance and speed in visually rich document retrieval tasks.
Weixin Liang, Lili Yu, Liang Luo +8 authors
Mixture-of-Transformers (MoT) reduces pretraining computational costs for multi-modal large language models by decoupling non-embedding parameters by modality, enabling efficient processing and competitive performance across text, image, and speech.
Frank F. Xu, Yufan Song, Boxuan Li +18 authors
A benchmark platform evaluates AI agents' autonomy in performing professional tasks in a simulated workplace environment, showing that a significant portion of simpler tasks can be automated but complex, long-term tasks remain challenging.
Jinheng Xie, Weijia Mao, Zechen Bai +7 authors
Show-o, a unified transformer model, combines autoregressive and diffusion modeling for versatile vision-language tasks, achieving competitive performance.
Mingan Lin, Fan Yang, Yanjun Shen +22 authors
Baichuan Alignment provides comprehensive insights into alignment methodologies used in Baichuan models, detailing improvements through Prompt Augmentation System, Supervised Fine-Tuning, and Preference Alignment across various benchmarks.
Zhaorun Chen, Zhen Xiang, Chaowei Xiao +2 authors
AgentPoison, a novel red teaming approach, introduces a backdoor attack targeting LLM agents by poisoning their memory or knowledge base, ensuring high attack success rates with minimal impact on normal performance.
Siyu Zhou, Tianyi Zhou, Yijun Yang +4 authors
A neurosymbolic approach aligns large language models with their environment through rule learning, improving model-based agent performance in open-world tasks.
北京市昌平区探索星信息技术及软件开发工作室
京ICP备2026059466号