Service and Contributions

Open Source Contributions

I have contributed to several open-source projects that advance Natural Language Processing (NLP) and Machine Learning (ML):

• Starcoder2 Project
  • Contributed to developing Starcoder2, an open-source coding LLM, as part of the BigCode project led by Hugging Face.
  • Impact: Recognized as one of the leading open-source LLMs for coding in 2024, with complete transparency in its training process and data usage.
  • GitHub Repository
• FACTUAL: Text Scene Graph Parsing
  • Built FACTUAL, a text scene graph parser, in collaboration with Adobe and Wuhan University.
  • Impact: Downloaded over 33,000 times from PyPI and 130,000 times from Hugging Face Models.
  • GitHub Repository
• SCAR: Data Selection Tool for LLM Fine-tuning
  • Developed SCAR, a state-of-the-art tool for data selection that enhances LLM performance while reducing dataset size.
  • Impact: Downloaded over 5,000 times from PyPI.
  • GitHub Repository

Peer Review Activities

I contribute to the academic community by reviewing papers for top-tier conferences and journals:

ACL, EMNLP, ICLR, EACL, NAACL, ACL Rolling Review, AAAI, AJI

Talks and Workshops

I have delivered talks and organized workshops to share insights on NLP and LLM research with academic and industry audiences:

2024

• Detect Automatic AI-Generated Sentences for Human-AI Hybrid Articles
  • Event: Shared Task at ALTA 2024, Canberra
  • Details
• Make Sense of Textual Data
  • Event: ResBaz Workshop, Victoria
  • Details
• Synthetic or Human Data: Optimizing Data Curation for Alignment of Large Language Models
  • Event: Virtual Talk at Ant Group
  • Details

2020

• Context-Dependent Semantic Parsing
  • Event: Conference Oral Presentation at COLING 2020
  • Details