Gang Liu
@gliu0329
Ph.D. student @ND_CSE | Graph ML | AI for Chemistry and Material Science | Research Intern @Amazon, @broadinstitute, @MITIBMLab
Introducing🔥torch-molecule🔥: A single line of code for molecular property prediction, generation & representation learning: > 30 deep learning methods + models, sklearn-style. All available at: `pip install torch-molecule` Code: github.com/liugangcode/to…
Introducing DeepEvolve 🧠: a research & code agent powered by Deep Research 🔍 and AlphaEvolve ⚙️ to discover new ideas/algorithms across scientific domains: Code: github.com/liugangcode/de… Support multi-file code evolution, auto debugging, and broader domains (math, bio, chem).
Join The Open Polymer Challenge! A significant barrier to AI-driven polymeric material discovery is data accessibility. Unlike computer vision or language, where massive open datasets enabled breakthrough progress, there has been relatively little polymer data despite the…
Nearly ~1k submissions to the kaggle competition and it just started! 3 months left!
🎓💰🔬 Want to learn machine learning, win a cash prize (USD 50K in total!!), and help drive real progress in discovering new polymer materials? All available at our NeurIPS 2025 Open Polymer Challenge: open-polymer-challenge.github.io 🚀 Join now (Kaggle): kaggle.com/competitions/n… 📈
Join our NeurIPS competition on Kaggle: learn machine learning, make a scientific impact, and earn prize money. 🥳
📣 Competition Launch Alert! NeurIPS Open Polymer Prediction 2025 🎯 Develop ML algorithms to predict polymer properties 💰 $50,000 Prize Pool ⏰ Entry Deadline: September 8, 2025 🙏 @NeurIPSConf, @ND_CSE and @gliu0329 kaggle.com/competitions/n…
Open Polymer Challenge: Leveraging Machine Learning for Polymer Informatics was accepted to NeurIPS 2025 Competition Track and is now LAUNCHed on Kaggle! JOIN US AND WIN $50,000 Awards! YES, FOUR "0"s - it's $50,000! Soooo what are YOU waiting for???
🎓💰🔬 Want to learn machine learning, win a cash prize (USD 50K in total!!), and help drive real progress in discovering new polymer materials? All available at our NeurIPS 2025 Open Polymer Challenge: open-polymer-challenge.github.io 🚀 Join now (Kaggle): kaggle.com/competitions/n… 📈
#Transformers are all you need to generate graphs. In my coming @icmlconf paper, I show how to convert every graph to a unique sequence of transformation rules. Treating rules like words, #LLMs will eat up graphs just like sentences! Read more: arxiv.org/abs/2505.22949
MolTextNet: A Two-Million Molecule-Text Dataset for Multimodal Molecular Learning １．MolTextNet introduces a dataset of 2.5 million molecule-text pairs, designed to support large-scale multimodal learning between molecular graphs and natural language. Each entry provides…
So many things in the run-up to DSPy 3. Here's a first, EXPERIMENTAL one: 🚨We're releasing dspy.GRPO, an online RL optimizer for DSPy programs Your DSPy code as-is can be dspy.GRPO'ed. Yes, even compound multi-module programs. Led by @NoahZiems @LakshyAAAgrawal @dilarafsoylu.
Excited to be in Singapore for @iclr_conf from April 24-28! We’re presenting two posters on multimodal molecular learning—come say hello if you’d like to chat about our work or related topics. Looking forward to great discussions!
We were so happy having Prof. Xifeng Yan (UCSB) at Notre Dame today. He chatted with our @ND_CSE graduate students, met many old and new friends, and gave a wonderful talk about his recent thoughts on language models. Thanks, Xifeng! 😃
Can LLMs help us discover the right molecules for new medicines? An MIT-led approach combines an LLM w/graph-based models to create new molecules w/specific properties. Users can request molecules in plain language & get a description of how to make it: bit.ly/3RerIZY
📣Congrats to Gang Liu, a doctoral student in @ND_CSE, on receiving a 2024-25 IBM PhD Fellowship! Liu’s research applies graph machine learning to discover new materials. Congratulations!🎉🎉 engineering.nd.edu/news/gang-liu-…
Our lab computation (3*NVDA a6k 48GB per server) can only do GRPO training with 1.5B size models, which seems to not work with such a small model. 😅 Any recommendation on computation-efficient training framework (probably train a 3B model with GRPO)?