A5107249140- Software Engineering Research
- Model-Driven Software Engineering Techniques
- Probability and Risk Models
- Scientific Computing and Data Management
- Parallel Computing and Optimization Techniques
- Software Testing and Debugging Techniques
- Advanced Measurement and Metrology Techniques
- Natural Language Processing Techniques
- Topic Modeling
- Speech Recognition and Synthesis
- Manufacturing Process and Optimization
- Advanced Harmonic Analysis Research
- Statistical Distribution Estimation and Applications
- Random Matrices and Applications
- Stochastic processes and statistical mechanics
Anhui University
2019-2023
ML-powered code generation aims to assist developers write in a more productive manner by intelligently generating blocks based on natural language prompts. Recently, large pretrained deep learning models have pushed the boundary of and achieved impressive performance. However, huge number model parameters poses significant challenge their adoption typical software development environment, where developer might use standard laptop or mid-size server develop code. Such cost resources terms...
Hantian Ding, Varun Kumar, Yuchen Tian, Zijian Wang, Rob Kwiatkowski, Xiaopeng Li, Murali Krishna Ramanathan, Baishakhi Ray, Parminder Bhatia, Sudipta Sengupta. Proceedings of the 61st Annual Meeting Association for Computational Linguistics (Volume 5: Industry Track). 2023.
Code generation models are not robust to small perturbations, which often lead inconsistent and incorrect generations significantly degrade the performance of these models. Improving robustness code is crucial better user experience when deployed in real-world applications. However, existing efforts have addressed this issue for To fill gap, we propose CodeFort, a framework improve models, generalizing large variety perturbations enrich training data enabling various strategies, mixing...
Fill-in-the-Middle (FIM) has become integral to code language models, enabling generation of missing given both left and right contexts. However, the current FIM training paradigm, which reorders original sequences then performs regular next-token prediction (NTP), often leads models struggling generate content that aligns smoothly with surrounding context. Crucially, while existing works rely on rule-based post-processing circumvent this weakness, such methods are not practically usable in...
In this paper, the complete moment convergence for arrays of rowwise negatively superadditive dependent (NSD, short) random variables is established. As applications, and Marcinkiewicz-Zygmund type strong law large numbers NSD are also obtained. Finally, a numerical simulation carried out to verify validity theoretical results. The results obtained in paper extend corresponding ones literature.
ML-powered code generation aims to assist developers write in a more productive manner, by intelligently generating blocks based on natural language prompts. Recently, large pretrained deep learning models have substantially pushed the boundary of and achieved impressive performance. Despite their great power, huge number model parameters poses significant threat adapting them regular software development environment, where developer might use standard laptop or mid-size server develop her...
AbstractIn this article, the Marcinkiewicz-Zygmund-type moment inequality and Rosenthal-type for m-extended negatively dependent (m-END, short) random variables are established. As applications of inequality, we further investigate Lr convergence properties arrays rowrise m-END variables. Some sufficient conditions provided. Finally, some simulations presented to verify validity theoretical results. The results obtained in article generalize known ones independent variables.Keywords:...