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。
Albert Gu, Tri Dao
Mamba, a novel SSM-based model, outperforms Transformers in inference speed and scalability across various modalities by selectively propagating information and using efficient hardware-aware algorithms.
50 篇论文 · 按点赞排序
Yuxiang Wei, Zhe Wang, Jiawei Liu +2 authors
Magicoder, using OSS-Instruct to incorporate open-source code snippets, achieves superior performance on coding benchmarks while reducing bias in synthetic data generation.
Zhen Li, Mingdeng Cao, Xintao Wang +3 authors
PhotoMaker is an efficient personalized text-to-image generation method that encodes ID images into a unified embedding, achieving high ID fidelity, speed improvements, and generalization capabilities with an ID-oriented data pipeline.
Vladimir Arkhipkin, Andrei Filatov, Viacheslav Vasilev +6 authors
Kandinsky 3.0, a large-scale text-to-image model based on latent diffusion, improves quality and realism through a larger architecture and advanced text understanding.
Hyeonho Jeong, Geon Yeong Park, Jong Chul Ye
The VMC framework enhances video generation by tuning temporal attention layers in diffusion models to accurately reproduce and diversify motion with a motion distillation objective.
Zehua Chen, Guande He, Kaiwen Zheng +2 authors
Bridge-TTS introduces a novel text-to-speech system using a clean deterministic prior in diffusion models, significantly improving synthesis quality and sampling efficiency over existing methods.
Chengshu Li, Jacky Liang, Andy Zeng +7 authors
Chain of Code (CoC) enhances language models' reasoning capabilities by using pseudocode and selective code emulation, enabling them to tackle complex semantic and logical tasks more effectively.
Zeyi Sun, Ye Fang, Tong Wu +6 authors
Alpha-CLIP enhances CLIP by adding an auxiliary alpha channel for attentive region suggestion, enabling precise control over image content across various tasks like open-world recognition and multimodal generation.
Peng Wang, Yichun Shi
ImageDream is a multi-view diffusion model for 3D object generation that uses image prompts to produce high-quality 3D models with improved visual geometry accuracy.
Tero Karras, Miika Aittala, Jaakko Lehtinen +3 authors
Modifications to network layers in the ADM diffusion model architecture improve training stability and synthesis quality, reducing FID from 2.41 to 1.81, and a novel method for post-hoc EMA parameter tuning is introduced.
Yuxuan Yan, Chi Zhang, Rui Wang +3 authors
A hybrid guidance framework for identity-preserving image synthesis efficiently generates stylistic portraits by combining stylized images, facial images, and textual prompts.
Shunsuke Saito, Gabriel Schwartz, Tomas Simon +2 authors
Relightable Gaussian Codec Avatars model high-fidelity head avatars with real-time relighting capabilities using 3D Gaussians for geometry and learnable radiance transfer for appearance.
Bill Yuchen Lin, Abhilasha Ravichander, Ximing Lu +5 authors
A study explores the effectiveness of alignment tuning in large language models, suggesting that superficial methods like in-context learning can achieve strong performance without traditional supervised fine-tuning or reinforcement learning.
Yuelang Xu, Benwang Chen, Zhe Li +4 authors
The method uses controllable 3D Gaussians and a fully learned MLP-based deformation field for high-fidelity 3D head avatar modeling under sparse views, employing geometry-guided initialization with implicit SDF and Deep Marching Tetrahedra for stability and high rendering quality.
Lingmin Ran, Xiaodong Cun, JiaWei Liu +5 authors
X-Adapter upgrades text-to-image diffusion models like SDXL to work with existing plugins without retraining, through feature remapping and a null-text training strategy.
En Yu, Liang Zhao, Yana Wei +8 authors
A novel MLLM, Merlin, is introduced which incorporates future modeling through FPT and FIT, enhancing foresight and future reasoning capabilities.
Xuan-Phi Nguyen, Wenxuan Zhang, Xin Li +9 authors
SeaLLMs, a series of Llama-2 based models tailored for Southeast Asian languages, demonstrate superior performance across linguistic tasks and outperform ChatGPT-3.5 in non-Latin languages while being lightweight and cost-effective.
Jiaman Li, Alexander Clegg, Roozbeh Mottaghi +3 authors
CHOIS uses a conditional diffusion model to generate synchronized human and object motion in 3D scenes, guided by language descriptions and waypoints, with improvements for object geometry alignment and contact constraints.
Jiaming Han, Kaixiong Gong, Yiyuan Zhang +6 authors
A unified framework aligns multiple modalities to language using OneLLM, a multimodal large language model, demonstrating strong performance across various tasks.
Xinyin Ma, Gongfan Fang, Xinchao Wang
DeepCache, a training-free method, accelerates diffusion models by caching and reusing features across denoising stages, achieving significant speedups with minimal quality degradation.
Yuming Jiang, Tianxing Wu, Shuai Yang +5 authors
VideoBooth generates high-quality customized videos using image prompts through a coarse-to-fine embedding and attention injection module, achieving state-of-the-art results in video content creation.
Tong Wu, Zhibing Li, Shuai Yang +5 authors
HyperDreamer enables realistic and editable 3D modeling from a single image using viewable 360-degree meshes, renderable materials, and text-guided editing.
Yuchao Gu, Yipin Zhou, Bichen Wu +7 authors
The VideoSwap framework uses semantic point correspondences for video subject swapping, achieving superior results in handling shape changes compared to dense correspondence methods.
Xiao Ma, Swaroop Mishra, Ariel Liu +6 authors
ExploreLLM facilitates exploratory tasks and personalization by integrating natural language with structured interfaces in LLM-powered chatbots.
Zhouxia Wang, Ziyang Yuan, Xintao Wang +4 authors
MotionCtrl is a unified motion controller for video generation that independently controls camera and object motion, offering more flexible and diverse video outputs.
Xiaoke Huang, Jianfeng Wang, Yansong Tang +5 authors
A method enhances the Segment Anything Model (SAM) with regional caption generation using a lightweight query-based feature mixer, enabling efficient use of weak supervision pretraining on publicly available datasets.
Felix Wimbauer, Bichen Wu, Edgar Schoenfeld +11 authors
Block caching in diffusion models improves image generation quality without increasing computational cost by recycling layer outputs.
Nuo Chen, Ning Wu, Shining Liang +4 authors
An analysis of LLaMA focusing on reasoning and computation through multiple-choice tasks reveals layer-specific knowledge and computational capabilities without significant enhancement from larger model sizes.
Jiwen Yu, Xiaodong Cun, Chenyang Qi +4 authors
AnimateZero enhances pre-trained text-to-video diffusion models with precise appearance and motion control capabilities, enabling finer control over the generation process and supporting new applications like interactive video generation.
Xi Chen, Zhiheng Liu, Mengting Chen +4 authors
A text-to-video system, LivePhoto, uses a text-to-image generator enhanced with a motion module to accurately translate textual motion descriptions into videos.
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