Lichang Chen
@LichangChen2
Followers
653
Following
979
Statuses
235
AI/ML PhDing @UMDCS | GenAI Unit Intern @GoogleDeepmind| ex @NVIDIA @GoogleAI | Building the AGI | BS @ZJU_China | Opinions are my own.
College Park, MD
Joined September 2021
Some life updates: I will have a back-to-back intern @GoogleDeepMind , and join @kelvin_guu’s team as SR next week. During the past summer, the most memorable moment for me is the $50 bet with Pranav @recurseparadox: If the better format of synthetic data can win over 10%. I bet “less” but the results surprised me😳😳: a small change of the format can bring 9% improvement. I won the bet but I really admired the insights and good sense Pranav has. I am so proud of being part of the GDM team and hope to do more impactful work in the next 6 months!
5
4
161
RT @umdcs: Honoring Qiang Yang, a leading researcher in AI and data mining. #UMDCS #AlumniHallOfFame
0
1
0
RT @melvinjohnsonp: Big Gemini 2.0 news today. - 2.0 Pro is now available as experimental and is #1 across all categories. - 2.0 Flash is…
0
17
0
RT @shaneguML: Who knew Software 2.0 wasn't the moat, but Software 1.0 is. Google fortunately has both. Google seems well prepared for mul…
0
2
0
@hexianghu Thanks, Frank! I am grateful that I learned a lot from you more than I expected! Super great intern experience!
0
0
1
Thrilled to announce that two papers I was done @GoogleDeepMind are accepted by #ICLR2025! 🎉🎉 One is about reasoning on Omni-modality Language Model evaluations and one is for robust reward model training to reduce hackings on RLHF. Great thanks to my collaborators!
0
4
49
RT @srush_nlp: 10 short videos about LLM infrastructure to help you appreciate Pages 12-18 of the DeepSeek-v3 paper (
0
121
0
RT @melvinjohnsonp: Gemini-Exp-1206 is now available on the Gemini app. It’s the #1 model on LMArena across all categories and people seem…
0
6
0
1
0
3
Go and work with Xi!
🔔 I'm recruiting multiple fully funded MSc/PhD students @UAlberta for Fall 2025! Join my lab working on NLP, especially reasoning and interpretability (see my website for more details about my research). Apply by December 15!
0
0
2
@arvind_io @manohar_paluri @Ahmad_Al_Dahle @edunov @AIatMeta Congrats! BTW: Could you reply my message?
0
0
0
RT @IHung_Hsu: Our team (Google Cloud AI Research: is seeking PhD student researchers/interns to work on LLM-relat…
0
83
0
@jxmnop Here are several reasons I think why it is hard to replicate: 1. We do not have very high-quality labelled IMO/IOI data 2. Not enough compute since inference time RL is also compute-intensive task. 3. Base Model not strong enough for exploring the inference-time algorithms
0
0
4