Trends · 研究趋势
数据来自 Hugging Face 论文的 AI 提取关键词,按月统计研究方向的增长与热度。
Shenghai Yuan, Yuanyang Yin, Zongjian Li +3 authors
Helios is a 14 billion parameter autoregressive diffusion model for video generation that achieves real-time performance and high-quality long-video synthesis without conventional optimization techniques.
Team Seedream, Yunpeng Chen, Yu Gao +47 authors
Seedream 4.0 is a high-performance multimodal image generation system that integrates text-to-image synthesis, image editing, and multi-image composition using a diffusion transformer and VAE, achieving state-of-the-art results with efficient training and inference.
Andrey Gromov, Kushal Tirumala, Hassan Shapourian +2 authors
Layer pruning of pre-trained LLMs with parameter-efficient finetuning methods shows minimal performance degradation and significant resource savings in both finetuning and inference.
Tianchen Zhao, Ke Hong, Xinhao Yang +8 authors
PAROAttention reorganizes visual attention patterns to enable efficient sparsification and quantization, reducing memory and computational costs with minimal impact on performance.
Jintao Zhang, Haofeng Huang, Pengle Zhang +3 authors
SageAttention2 improves attention computation speed with 4-bit matrix multiplication and adaptive quantization, achieving significant OPS increases without compromising accuracy.
Igor Pavlovic, Thiemo Wandel, Anton Obukhov +6 authors
Marigold V2 repurposes diffusion transformers for monocular depth estimation via single-step flow-matching inference, semantic alignment, and a Sinkhorn-based two-stage fine-tuning protocol, yielding sharper out-of-distribution depth maps and strong results on related dense regression tasks.
Jintao Zhang, Jia wei, Pengle Zhang +2 authors
A new quantization method for attention mechanisms, named SageAttention, enhances inference speed and accuracy without significant loss in performance across various models.
Abdul Fatir Ansari, Lorenzo Stella, Caner Turkmen +14 authors
Chronos, a framework using pretrained transformer-based models for time series forecasting, outperforms classical methods and achieves comparable zero-shot performance on unseen datasets using tokenized time series data.
Moo Jin Kim, Karl Pertsch, Siddharth Karamcheti +15 authors
OpenVLA, a 7B-parameter open-source vision-language-action model, demonstrates strong performance in generalist manipulation and efficient fine-tuning for new tasks, outperforming larger closed models and from-scratch imitation learning methods.
Yuhui Xu, Lingxi Xie, Xiaotao Gu +6 authors
QA-LoRA algorithm reduces computational requirements for large language models by quantizing weights during fine-tuning and integrating auxiliary weights post-training.
Donghyun Lee, Jitesh Chavan, Duy Nguyen +5 authors
OrbitQuant enables efficient post-training quantization for diffusion transformers by using a normalized rotated basis that eliminates the need for recalibration across different timesteps and modalities.
Yixiao Li, Yifan Yu, Chen Liang +4 authors
LoftQ, a novel quantization framework, improves performance of LLMs by simultaneously quantizing and initializing LoRA fine-tuning, reducing the gap in downstream task performance.
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