no code implementations • 29 Jan 2024 • Jian Gu, Aldeida Aleti, Chunyang Chen, Hongyu Zhang
In response, we introduce a pioneering method called vocabulary-defined semantics, which establishes a reference frame within the LM latent space, ensuring disentangled semantic analysis grounded in LM vocabulary.
no code implementations • 8 Dec 2023 • Jian Gu, Aldeida Aleti, Chunyang Chen, Hongyu Zhang
In this paper, we propose a novel and effective model editing approach, \textsc{MENT}, to patch LLMs in coding tasks.
no code implementations • 4 Sep 2022 • Jian Gu, Harald C. Gall
To systematically explore the potential improvements of code generation, we let it participate in the whole top-down development from \emph{expressibles} to \emph{executables}, which is possible in limited scopes.
1 code implementation • 29 Mar 2022 • Jian Cheng, Yanguang Wan, Dexin Zuo, Cuixia Ma, Jian Gu, Ping Tan, Hongan Wang, Xiaoming Deng, yinda zhang
3D hand pose estimation from single depth is a fundamental problem in computer vision, and has wide applications. However, the existing methods still can not achieve satisfactory hand pose estimation results due to view variation and occlusion of human hand.
Ranked #1 on Hand Pose Estimation on ICVL Hands
1 code implementation • 13 Jan 2022 • Jian Gu, Pasquale Salza, Harald C. Gall
Thereby, we propose a flexible and robust approach for automatic code summarization, based on neural models.
Ranked #1 on Source Code Summarization on ParallelCorpus-Python
1 code implementation • 2 Jul 2021 • Jian Gu, Zimin Chen, Martin Monperrus
In this paper, to improve the vector space, we introduce tree-serialization methods on a simplified form of AST and build the multimodal representation for the code data.
Ranked #1 on Code Search on CodeSearchNet - Ruby
no code implementations • ECCV 2020 • Boren Li, Po-Yu Zhuang, Jian Gu, Mingyang Li, Ping Tan
As for the proposed method, we first train a foreground encoder to learn representations of interchangeable foregrounds.
no code implementations • 16 Jan 2020 • Hongwei Xie, Jiafang Wang, Baitao Shao, Jian Gu, Mingyang Li
Finally, we provide a variety of experimental results to show that the proposed framework is able to achieve state-of-the-art accuracy with significantly reduced computational cost, which are the key properties for enabling real-time applications in low-cost commercial devices such as mobile devices and AR/VR headsets.
no code implementations • ICCV 2019 • Boren Li, Boyu Zhuang, Mingyang Li, Jian Gu
The framework, called Seq-SG2SL, derives sequence proxies for the two modality and a Transformer-based seq-to-seq model learns to transduce one into the other.