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
Akshay Krishnamurthy, Keegan Harris, Dylan J. Foster +2 authors
Experimentation with LLMs in multi-armed bandit environments reveals that robust exploration requires specific prompts, external summarization of history, or algorithmic interventions.
Payel Das, Subhajit Chaudhury, Elliot Nelson +10 authors
Larimar, a brain-inspired architecture with distributed episodic memory, enhances LLMs for efficient, fast, and accurate knowledge updates, fact editing, and context length generalization.
Hugh Zhang, Jeff Da, Dean Lee +12 authors
Evaluation of large language models on a new math benchmark, GSM1k, reveals that many models exhibit signs of overfitting to the existing GSM8k benchmark, leading to performance drops on GSM1k.
William Brandon, Mayank Mishra, Aniruddha Nrusimha +2 authors
Cross-Layer Attention modifies transformer-based autoregressive large language models to reduce KV cache size while maintaining accuracy, enabling longer sequence lengths and larger batch sizes during inference.
Zipeng Fu, Tony Z. Zhao, Chelsea Finn
Co-training of supervised behavior cloning with static and mobile manipulation datasets improves the success rates of mobile manipulation tasks using a whole-body teleoperation system.
Zhennan Wu, Yang Li, Han Yan +8 authors
BlockFusion, a diffusion-based model, generates and expands 3D scenes using latent tri-plane representations and a 2D layout conditioning mechanism.
Haotian Zhang, Haoxuan You, Philipp Dufter +8 authors
Ferret-v2 improves on Ferret by incorporating flexible high-resolution grounding, multi-granularity visual encoding with DINOv2, and a three-stage training paradigm, leading to enhanced visual understanding and performance.
Shenghai Yuan, Jinfa Huang, Yujun Shi +6 authors
MagicTime enhances Text-to-Video synthesis by integrating real-world physics knowledge through metamorphic time-lapse videos, using the MagicAdapter, Dynamic Frames Extraction, and Magic Text-Encoder techniques.
Ruibo Liu, Jerry Wei, Fangyu Liu +8 authors
The success of AI models relies on the availability of large, diverse, and high-quality datasets, which can be challenging to obtain due to data scarcity, privacy concerns, and high costs. Synthetic data has emerged as a promising solution by generating artificial data that mimics real-world patterns. This paper provides an overview of synthetic data research, discussing its applications, challenges, and future directions. We present empirical evidence from prior art to demonstrate its effectiveness and highlight the importance of ensuring its factuality, fidelity, and unbiasedness. We emphasize the need for responsible use of synthetic data to build more powerful, inclusive, and trustworthy language models.
Mirac Suzgun, Adam Tauman Kalai
Meta-prompting enhances language model performance by organizing and integrating multiple expert instances to handle complex tasks, demonstrating superiority over traditional methods in zero-shot settings.
Zhen Ye, Zeqian Ju, Haohe Liu +10 authors
FlashSpeech, using latent consistency model and adversarial consistency training, achieves fast and high-quality zero-shot speech synthesis with enhanced prosody generation.
Anwen Hu, Haiyang Xu, Jiabo Ye +8 authors
A Unified Structure Learning framework is introduced to enhance Visual Document Understanding by integrating structure-aware parsing and multi-grained text localization tasks using a vision-to-text module H-Reducer, achieving state-of-the-art results on multiple benchmarks.
Yixing Jiang, Jeremy Irvin, Ji Hun Wang +3 authors
Multimodal foundation models exhibit significant performance improvements with many-shot in-context learning compared to few-shot, with Gemini 1.5 Pro showing higher data efficiency and better batch processing capabilities across various domains.
Oleksiy Ostapenko, Zhan Su, Edoardo Maria Ponti +5 authors
A novel approach using model-based clustering and zero-shot routing improves the generalization of adapters in large language models for new tasks.
Tianhe Ren, Qing Jiang, Shilong Liu +13 authors
Grounding DINO 1.5, comprising Pro and Edge models, enhances open-set object detection through advanced architectures and large datasets, achieving high performance and efficiency.
Chuofan Ma, Yi Jiang, Jiannan Wu +2 authors
Groma, a Multimodal Large Language Model, utilizes region-level visual tokenization to enhance visual understanding and localization accuracy in image-grounded tasks.
Xiaoyi Dong, Pan Zhang, Yuhang Zang +21 authors
InternLM-XComposer2-4KHD enhances high-resolution understanding capabilities in LVLMs up to 4K, surpassing existing models with dynamic patch configuration and consistent performance improvements.
Gerald Shen, Zhilin Wang, Olivier Delalleau +10 authors
NeMo-Aligner is a toolkit for aligning large language models with human values using scalable techniques like RLHF, DPO, SteerLM, and SPIN, optimized for use with hundreds of GPUs and parameter-efficient fine-tuning.
Sheng-Chieh Lin, Luyu Gao, Barlas Oguz +4 authors
Proposes factuality-aware alignment methods for pre-trained language models to enhance factual accuracy during supervised fine-tuning and reinforcement learning, reducing hallucination.
Eloi Alonso, Adam Jelley, Vincent Micheli +4 authors
DIAMOND, a reinforcement learning agent using a diffusion world model, achieves top performance on Atari 100k by capturing visual details effectively.
Navve Wasserman, Noam Rotstein, Roy Ganz +1 authors
A diffusion model is trained to add objects to images based on text descriptions using a large-scale dataset of images with removed objects, outperforming existing methods.
Yue Wu, Zhiqing Sun, Huizhuo Yuan +3 authors
A self-play method called SPPO for language model alignment achieves state-of-the-art performance by approximating Nash equilibrium policy in a constant-sum game setting, outperforming other approaches with limited data.
Octo Model Team, Dibya Ghosh, Homer Walke +15 authors
Octo, a large transformer-based policy trained on extensive robotic datasets, demonstrates versatility and efficient fine-tuning for diverse robotic platforms and tasks.
Zhenwei Shao, Zhou Yu, Jun Yu +5 authors
A systematic study of lightweight large multimodal models (LMMs) led to the development of the Imp family, which achieves superior performance compared to larger models and is capable of high-speed inference on mobile devices.
Wenyu Du, Tongxu Luo, Zihan Qiu +5 authors
A depthwise stacking operator called $G_{\text{stack}}$ accelerates LLM pre-training, improving performance with less computational expense and providing guidelines for effective use.
Siwei Wang, Yifei Shen, Shi Feng +3 authors
Theory and experiments show that Transformer-based models, while capable of learning path-finding through embedding adjacency matrices, struggle with identifying reachability through transitivity in network planning tasks.
Aaron Defazio, Xingyu, Yang +4 authors
A new schedule-free method achieves state-of-the-art performance in optimization without requiring a stopping time or additional hyper-parameters, by unifying scheduling and iterate averaging.
Shenghai Yuan, Jinfa Huang, Yongqi Xu +7 authors
ChronoMagic-Bench evaluates text-to-video models' ability to generate time-lapse videos with metamorphic amplitude and temporal coherence using novel metrics and a large dataset.
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