Stephen Lu
@stephenzlu
Followers
83
Following
300
Media
1
Statuses
19
🇨🇦 | ai for drug discovery | soon phd @berkeley_ai | prev @bighatbio, @mcgillu
Berkeley, California
Joined December 2019
RT @AndreiRekesh: 📢We are excited to share SynCoGen—the first generative model that co-generates 🔷building-block graphs,🔷reaction edges and….
0
24
0
How close are we to an AI scientist that can make novel biological discoveries? Our recent work benchmarks key experimental abilities of frontier LLMs using insights from systems biology. We find that agents struggle at manipulating complex systems at scale ⬇️.
What makes a great scientist? Most AI scientist benchmarks miss the key skill: designing and analyzing experiments. 🧪 We're introducing SciGym: the first simulated lab environment to benchmark #LLM on experimental design and analysis capabilities. #AI4SCIENCE #ICML25
0
0
4
Original paper: My codebase: Thanks for the awesome work! @HavasiMarton @ItaiGat @RickyTQChen (2/2).
github.com
Educational Implementation of "Edit Flows: Flow Matching with Edit Operations" by Havasi et al. - TheMatrixMaster/edit-flows-demo
0
0
2
🚀 Just implemented Edit Flows => flow matching via edit operations (insert/delete/substitute)! No official code yet, so I built my own educational version on discretized sine functions. Watching the rates converge during training is very satisfying! (1/2)
Padding in our non-AR sequence models? Yuck. 🙅. 👉 Instead of unmasking, our new work *Edit Flows* perform iterative refinements via position-relative inserts and deletes, operations naturally suited for variable-length sequence generation. Easily better than using mask tokens.
4
5
44
RT @GabriCorso: Excited to unveil Boltz-2, our new model capable not only of predicting structures but also binding affinities! Boltz-2 is….
0
419
0
RT @NikoMcCarty: We made a huge poster that illustrates all of the major genome editing tools in one place. You can download a copy for fr….
0
176
0
RT @BiologyAIDaily: Aligning Protein Conformation Ensemble Generation with Physical Feedback. １．This work introduces Energy-based Alignment….
0
6
0
RT @saagnikkk: 🚨 Paper Alert: “RL Finetunes Small Subnetworks in Large Language Models”. From DeepSeek V3 Base to DeepSeek R1 Zero, a whopp….
0
126
0
RT @NotinPascal: Have we hit a "scaling wall" for protein language models? 🤔 Our latest ProteinGym v1.3 release suggests that for zero-shot….
0
20
0
RT @KevinKaichuang: Encode protein structures as a series of discrete tokens, train a language model, and sample protein structural conform….
0
24
0
RT @jiarlu: 🤔Excited about Protein + AI? Join our "AI for Protein Design" tutorial at #AAAI2025, with gorgeous @Oxer22 @divnori @WengongJin….
0
10
0
RT @andrewwhite01: I've been thinking about how reasoning models will change AI applied to science. The recent papers from Deepseek/AI2/M….
0
21
0
RT @ycombinator: YC W24's @clarum_ai helps PE firms close more deals by accelerating their due diligence. Firms can upload their files & qu….
ycombinator.com
Clarum helps private equity firms close more deals by conducting accelerated and reliable due diligence.
0
20
0
RT @OpenAI: Introducing Sora, our text-to-video model. Sora can create videos of up to 60 seconds featuring highly detailed scenes, comple….
0
31K
0
RT @mcgillu: They are already innovators in digital patient care, protective equipment production and mobile apps. Meet the 2021 Schulich L….
0
4
0