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。
Xian Li, Ping Yu, Chunting Zhou +5 authors
A scalable instruction-following language model is built using auto-labelling and iterative self-augmentation and self-curation, outperforming other LLaMa-based models on Alpaca.
28 篇论文 · 按点赞排序
Hu Ye, Jun Zhang, Sibo Liu +2 authors
IP-Adapter introduces a lightweight decoupled cross-attention mechanism to enhance pretrained text-to-image diffusion models with image prompt capabilities, enabling multimodal image generation and compatibility with existing tools.
Yangyi Huang, Hongwei Yi, Yuliang Xiu +4 authors
TeCH reconstructs high-fidelity 3D human models from single images using descriptive text prompts, a personalized Text-to-Image diffusion model, and a hybrid DMTet representation, outperforming existing methods.
Niklas Muennighoff, Qian Liu, Armel Zebaze +7 authors
Instruction tuning using Git commits improves performance on natural language and coding tasks compared to other benchmarks, with models achieving state-of-the-art results on expanded HumanEvalPack.
Tear Gosling, Alpin Dale, Yinhe Zheng
PIPPA, a community-driven partially-synthetic dataset, provides over a million utterances across numerous conversation sessions to enhance role-play and conversational AI research and development.
Hao Ouyang, Qiuyu Wang, Yuxi Xiao +6 authors
CoDeF represents videos using a canonical content field and a temporal deformation field, enabling the lifting of image processing algorithms to video processing with superior cross-frame consistency.
Xiaofei Wang, Manthan Thakker, Zhuo Chen +7 authors
SpeechX integrates neural codec language modeling and multi-task learning to enable versatile speech generation, transformation, and enhancement tasks in both clean and noisy conditions.
Cheng Li, Mingyang Zhang, Qiaozhu Mei +4 authors
A multistage and multitask framework using large language models enhances personalized text generation by improving retrieval, ranking, summarization, synthesis, and generation skills.
Binhui Liu, Xin Liu, Anbo Dai +3 authors
The dual-stream diffusion net (DSDN) enhances video consistency and smoothness in text-to-video generation by using separate content and motion diffusion streams with a cross-transformer interaction module and motion decomposer/combiner.
Ariel N. Lee, Cole J. Hunter, Nataniel Ruiz
A fine-tuned and merged family of large language models named Platypus, using LoRA modules and a curated dataset, achieves top performance on the Open LLM Leaderboard with reduced data and compute.
Aojun Zhou, Ke Wang, Zimu Lu +8 authors
A novel code-based self-verification method significantly improves math reasoning in GPT-4 Code Interpreter, enhancing accuracy on the MATH dataset.
Shengming Yin, Chenfei Wu, Jian Liang +4 authors
DragNUWA is a diffusion-based video generation model that integrates text, image, and trajectory controls for fine-grained and open-domain video creation.
Andrea Gesmundo, Kaitlin Maile
The research introduces methods to incrementally enlarge transformer-based neural networks while preserving their functionality, enabling efficient training of larger models.
Zhiwei Liu, Weiran Yao, Jianguo Zhang +12 authors
A comprehensive comparison and new orchestration strategy for LLM-augmented Autonomous Agents (LAAs) demonstrate their effectiveness in decision-making and multi-step reasoning tasks.
Jie Huang, Wei Ping, Peng Xu +3 authors
RAVEN enhances in-context learning in retrieval-augmented encoder-decoder models by combining masked and prefix language modeling, achieving superior performance with fewer parameters.
Nan Ding, Tomer Levinboim, Jialin Wu +2 authors
Theoretical analysis shows that prefix language models outperform causal language models in in-context learning by converging to optimal solutions, while causal models exhibit dynamics similar to online gradient descent.
Yan Tai, Weichen Fan, Zhao Zhang +3 authors
LCL, a method emphasizing causal reasoning in link-context learning, enhances MLLMs' ability to recognize novel concepts and images with fewer training samples.
Zhen Xu, Sida Peng, Chen Geng +5 authors
A method using Hierarchical Distance Query and sphere tracing for recovering animatable and relightable neural avatars from sparse or monocular videos under unknown lighting.
Zhouxia Wang, Jiawei Zhang, Tianshui Chen +2 authors
RestoreFormer++ improves blind face restoration by incorporating fully-spatial attention over multi-scale features and using a realistic degrading model to enhance realness, robustness, and generalization.
William Shen, Ge Yang, Alan Yu +3 authors
A method using distilled feature fields combines 3D geometry with 2D semantic features from CLIP to enable few-shot learning for robotic manipulation with generalization to new objects and expressions.
Cal Peyser, Zhong Meng, Ke Hu +5 authors
Joint speech-text encoders achieve consistent cross-modal representations without specific sequence-length treatment, improving downstream WER through consistency losses.
Haozhe Liu, Mingchen Zhuge, Bing Li +4 authors
A new method in deep reinforcement learning extracts important states from unannotated video data to improve policy learning and agent behavior.
Alexander Martin, Haitian Zheng, Jie An +1 authors
A new zero-shot image-to-image translation method using text-guided latent diffusion models is introduced to bridge large domain gaps in translating between skulls and living animals.
Sathiya Kumaran Mani, Yajie Zhou, Kevin Hsieh +3 authors
A natural-language-based system using large language models generates task-specific code for network management, improving accuracy and privacy.
Shaan Bijwadia, Shuo-yiin Chang, Weiran Wang +3 authors
Text injection using joint end-to-end and internal language model training enhances capitalization and turn-taking detection in automatic speech recognition systems.
Patrick Fernandes, Daniel Deutsch, Mara Finkelstein +7 authors
AutoMQM uses large language models to identify and categorize errors in machine translation, offering interpretability and improving performance over traditional score prediction.
Yonatan Bitton, Hritik Bansal, Jack Hessel +6 authors
VisIT-Bench evaluates instruction-following vision-language models across a variety of tasks through human-authored instructions and multimodal generation assessments.
Edith Tretschk, Vladislav Golyanik, Michael Zollhoefer +3 authors
SceNeRFlow reconstructs non-rigid time-consistent scenes using dynamic-NeRF with a decomposed deformation model to handle larger motions.
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