Jia Li



                    English name: Jia Li (♂)
                    Chinese name: 李佳

                    PhD Student
                    School of Computer Science
                    Peking University (PKU)
                    No.5 Yiheyuan Road, Haidian District, Beijing, China
                    Email: lijia AT stu DOT pku DOT edu DOT cn

                       



I am a doctoral student at the School of Computer Science, Peking University (PKU). My supervisor is Prof. Zhi Jin and Prof. Ge Li. I expect to graduate in July 2025.

My research interests mainly focus on Code Generation and Large Language Models (LLMs) for Code.

Welcome to pay attention to our released LLM - aiXcoder-7B: It surpasses existing LLMs of similar scales in code generation and code completion tasks.


News

  • October 18, 2024. We release the technical report of aiXcoder-7B, a lightweight and effective large language model for code completion. [Paper]
  • September 26, 2024. Our paper "EvoCodeBench: An Evolving Code Generation Benchmark with Domain-Specific Evaluations" has been accepted by NeurIPS 2024.
  • May 16, 2024. Our paper "DevEval: A Manually-Annotated Code Generation Benchmark Aligned with Real-World Code Repositories" has been accepted by ACL 2024.
  • April 15, 2024. Our paper "Exploring and Unleashing the Power of Large Language Models in Automated Code Translation" has been accepted by FSE 2024.
  • March 30, 2024. We propose an evolving code generation benchmark named EvoCodeBench for evaluating code generation models in repository-level code generation. [Paper], [Data]
  • January 30, 2024, Our paper "Deep Learning for Code Generation: A Survey" has been accepted by SCIS.
  • December 9, 2023. Our paper "Hot or Cold? Adaptive Temperature Sampling for Code Generation with Large Language Models" has been accepted by AAAI 2024.
  • October 9, 2023. Our paper "Poison Attack and Poison Detection on Deep Source Code Processing Models" has been accepted by TOSEM.
  • July 28, 2023. Our paper "ZC3 Zero-Shot Cross-Language Code Clone Detection" has been accepted by ASE 2023 Technical Research Track.
  • June 28, 2023. Our paper "CodeEditor: Learning to Edit Source Code with Pre-trained Models" has been accepted by ASE 2023 Journal-First Track.
  • May 29, 2023. Our paper "MCodeSearcher Multi-View Contrastive Learning for Code Search" has been accepted by Internetware 2023.
  • May 2, 2023. Our paper "Self-Edit: Fault-Aware Code Editor for Code Generation" has been accepted by ACL 2023.
  • April 7, 2023. Our paper "CodeEditor: Learning to Edit Source Code with Pre-trained Models" has been accepted by TOSEM.
  • December 9, 2022. Our paper "SkCoder: A Sketch-based Approach for Automatic Code Generation" has been accepted by ICSE 2023.
  • September 15, 2022. Our paper "Fine-Tuning Pre-Trained Language Models Effectively by Optimizing Subnetworks Adaptively" has been accepted by NeurIPS 2022.
  • July 8, 2021. Our paper "EditSum: A Retrieve-and-Edit Framework for Source Code Summarization" has been accepted by ASE 2021.