Qiguang Chen (陈麒光)

Hello! I’m a first-year master degree candidate of Harbin Institute of Technology, majoring in Computer Science, at Computing Faculty. My research interests includes Multi-modal, Multi-lingual and Chain-of-Thought Reasoning.

Currently, I am a master at SCIR supervised by Prof. Libo Qin and Prof. Wanxiang Che. Previously, as an undergraduate student, I have done research intern at Baidu NLP , supervised by Lijie Wang and Dr. Jing Liu.

Other than my work, I’m honored as Deputy Secretary-General for MLNLP. Previously, I was the president of HIT Software Competition Club from 2021 to 2023.

News Publications Honors&Awards Patent Applications

News

  • [2023.08] Our survey about LLM Competency is accepted by CCL 2023.
  • [2023.08] Our LLM (HuoZi) is officially released.
  • [2023.07] Our unified SLU toolkit (OpenSLU) is accepted by ACL 2023 (Demo).
  • [2023.07] Our works (CLIPText and MMSD2.0) are accepted by ACL 2023 (Findings).
  • [2023.06] Luckily, I win an outstanding graduate of HIT!
  • [2022.10] Fortunately, I win the CCF Excellent College Students Price!
  • [2022.10] Lead by Dr. Bo Zheng and Prof. Wanxiang Che, our paper achieves achieve the Best Paper in EMNLP MMNLU2022 WorkShop.
  • [2022.09] I formally join SCIR, HIT.
  • [2022.08] Our follow-up work on inconsistency in task-oriented dialogue systems is accepted by COLING2022. Thanks for co-authors from HIT, HKU, BUAA, CUMC! More exciting, with the guidance of Prof. Wanxiang Che, our team, lead by Dr. Bo Zheng, win the first price in MMNLU-22 Competition.
  • [2022.05] I join Baidu NLP, Baidu Inc.
  • [2022.04] Our work on multi-lingual SLU is accepted by ACL 2022. Thanks for co-authors from HIT, MSRA and NUS!
  • [2021.09] Our work on inconsistencies in Task-oriented Dialogue System is accepted by EMNLP 2021. Thanks for co-authors from HIT!

Publications

LLM SurveyThrough the Lens of Core Competency: Survey on Evaluation of Large Language Models
Ziyu Zhuang, Qiguang Chen, Longxuan Ma, Mingda Li, Yi Han, Yushan Qian, Haopeng Bai, Weinan Zhang, Ting Liu
[CCL 2023]
PDF

We summarize 4 core competencies of LLM, including reasoning, knowledge, reliability, and safety. For every competency, we introduce its definition, corresponding benchmarks,and metrics. Under this competency architecture, similar tasks are combined to reflect corresponding ability, while new tasks can also be easily added into the system. Finally, we give our suggestions on the future direction of LLM’s evaluation.

OpenSLUOpenSLU: A Unified, Modularized, and Extensible Toolkit for Spoken Language Understanding
Libo Qin*, Qiguang Chen*, Xiao Xu, Yunlong Feng, Wanxiang Che
[ACL 2023 (Demo)]
PDF / Code

In this work, we introduce OpenSLU, an open-source toolkit to provide a unified, modularized, and extensible toolkit for spoken language understanding. Specifically, OpenSLU unifies 10 SLU models for both single-intent and multi-intent scenarios, which support both non-pretrained and pretrained models simultaneously. Additionally, OpenSLU is highly modularized and extensible by decomposing the model architecture, inference, and learning process into reusable modules, which allows researchers to quickly set up SLU experiments with highly flexible configurations.

CLIPTextCLIPText: A New Paradigm for Zero-shot Text Classification
Libo Qin, Weiyun Wang, Qiguang Chen, Wanxiang Che
[ACL 2023 (Findings)]
PDF / Code

We introduce CLIPText, a novel paradigm for zero-shot text classification, which reformulates zero-shot text classification into a text-image matching problem that CLIP can be applied to. In addition, we further incorporate prompt into CLIPText (Prompt-CLIPText) to better derive knowledge from CLIP.

MMSD2.0MMSD2.0: Towards a Reliable Multi-modal Sarcasm Detection System
Libo Qin, Shijue Huang, Qiguang Chen, Chenran Cai, Yudi Zhang, Bin Liang, Wanxiang Che, Ruifeng Xu
[ACL 2023 (Findings)]
PDF / Code

We introduce MMSD2.0, a correction dataset that fixes the shortcomings of MMSD, by removing the spurious cues and re-annotating the unreasonable samples. Meanwhile, we present a novel framework called multi-view CLIP that is capable of leveraging multi-grained cues from multiple perspectives (i.e., text, image, and textimage interaction view) for multi-modal sarcasm detection.

MMNLU-22HIT-SCIR at MMNLU-22: Consistency Regularization for Multilingual Spoken Language Understanding
Bo Zheng, Zhouyang Li, Fuxuan Wei, Qiguang Chen, Libo Qin, Wanxiang Che.
[EMNLP 2022 MMNLU Workshop Best Paper]
PDF / Code

To improve the performance of these two sub-tasks, we propose to use consistency regularization based on a hybrid data augmentation strategy. The consistency regularization enforces the predicted distributions for an example and its semantically equivalent augmentation to be consistent.

CGIMCGIM: A Cycle Guided Interactive Learning Model for Consistency Identification in Task-oriented Dialogue
Libo Qin, Qiguang Chen, Tianbao Xie, Qian Liu, Shijue Huang, Wanxiang Che, Zhou Yu.
[COLING 2022]
PDF / Code

This work aims to solve CI-ToD task by introducing an explicit interaction paradigm, Cycle Guided Interactive learning Model (CGIM), which achieves to make information exchange explicitly from all the three tasks.

GLCLeFGL-CLeF: A Global-Local Contrastive Learning Framework for Cross-lingual Spoken Language Understanding.
Libo Qin, Qiguang Chen, Tianbao Xie, Qixin Li, Jian-Guang Lou, Wanxiang Che, Min-Yen Kan.
[ACL 2022]
PDF / Code

We present Global-Local Contrastive Learning Framework (GL-CLeF) to address this shortcoming. Specifically, we employ contrastive learning, leveraging bilingual dictionaries to construct multilingual views of the same utterance, then encourage their representations to be more similar than negative example pairs, which achieves to explicitly align representations of similar sentences across languages.

CIToDDon't be Contradicted with Anything! CI-ToD: Towards Benchmarking Consistency for Task-oriented Dialogue System.
Libo Qin, Tianbao Xie, Shijue Huang, Qiguang Chen, Xiao Xu, Wanxiang Che.
[EMNLP2021]
PDF / Code

We introduce CI-ToD, a novel dataset for Consistency Identification in Task-oriented Dialog system. In addition, we not only annotate the single label to enable the model to judge whether the system response is contradictory, but also provide more fine-grained labels (i.e., Dialogue History Inconsistency, User Query Inconsistency and Knowledge Base Inconsistency) to encourage model to know what inconsistent sources lead to it.

Honors&Awards

  • Bachelor of Engineering (B.E.)
    • Outstanding Graduates, 2023.
    • CCF Excellent College Students Price (Only 100 CS students in China were invited), 2022.
    • EMNLP MMNLU WorkShop Best Paper, 2022.
    • 1st place in MMNLU WorkShop Full-shot Setting, 2022.
    • National Second Prize in Contemporary Undergraduate Mathematical Contest in Modeling, China, 2021.
    • Provincial Gold Award in “Challenge” Cup, 2022.

Patent Applications

  • 预训练对偶注意力神经网络语义推断对话检索方法及系统、检索设备、存储介质。

    梁晨; 陈麒光; 耿健; 唐亚锋; 辛宇鑫

    ZL 2021 1 0795247.7, 2022.09.09.

Open Rescource Project

SimBiber: A tool for simplifying bibtex with official info.

version Status-building PRs-Welcome stars FORK Issues

Top-AI-Conferences-Paper-with-Code

version Status-building PRs-Welcome stars FORK Issues

AI-Paper-Collector

version Status-building PRs-Welcome stars FORK Issues

Trust-AI