Meng Jiang

🚨Typical RL algorithms and on-policy distillation methods are blind samplers: they use privileged info to score rollouts, but not to *find* them. We ask: can we use privileged info to *actively sample* the rollouts RL wishes it can stumble upon with compute? ⤵️ Pedagogical RL

I’m thrilled to share that I’ll be starting my CS PhD at @NorthwesternU this fall, advised by @ManlingLi_! I’ll be researching areas in trustworthy AI and spatial intelligence to build reliable AI systems that are grounded in the physical world. I’m also happy to announce that I was awarded the @NSF GRFP fellowship, which will support my PhD for 3 years! This wouldn’t have been possible without my wonderful mentors @nunompmoniz, @Meng_CS, @frank_liu_01, @NoahZiems, and countless others who’ve guided me throughout my undergrad. And so… I guess I won’t be leaving the midwest :)

Excited to share our ACL 2026 work, trying to solve the issue raised by the ICLR Outstanding Paper “LLMs Get Lost In Multi-Turn Conversation”! Our RLAAR (arxiv.org/pdf/2510.18731) is an RL framework that trains LLMs to both answer correctly and wait when context is insufficient, using verifiable accuracy and abstention rewards. This tackles a key weakness in today’s conversational LLMs: they often answer too early, make wrong assumptions, and struggle to recover as conversations unfold. We’re also excited to see this challenge highlighted by “LLMs Get Lost In Multi-Turn Conversation” (arxiv.org/pdf/2505.06120) being recognized as an ICLR 2026 Outstanding Paper. Reliable conversational AI needs to know when to answer — and when to hold back. #ACL2026 #ICLR2026 #LLM #RLVR #ConversationalAI

@lateinteraction Some supercomputer has 10M CPU cores; we human bodies don't even have five cores. Too painful! (well, sometimes, I feel some people have 120 hours/day.)

Being this excited about five rather unexpected research projects simultaneously is almost too painful. Assuming that we figure out how to sequence these releases, y’all are going to thoroughly love each of these.

Thrilled to present GEPA as an Oral Talk and Poster at ICLR 2026 this Friday in Rio! 🇧🇷 Apr 24 Oral Session 3A (Agents), 10:30 AM BRT, Amphitheater Poster Session 4, 3:15 PM, Pavilion 3 x.com/LakshyAAAgrawa… Let's recap what's happened since we released GEPA last year 🧵

@matei_zaharia @databricks Congratulations!

@databricks Definitely unexpected! It wouldn't have been possible without my collaborators at Databricks and my grad students.

We're incredibly proud to congratulate our co-founder and CTO, @matei_zaharia, on receiving the ACM Prize in Computing for his development of distributed data systems that have enabled large-scale machine learning, analytics, and AI. Matei's open-source contributions have fundamentally changed how organizations work with data and AI — including Apache Spark™, Delta Lake, and MLflow. Researchers, nonprofits, startups, and enterprises across every industry have built on the foundation he helped create. Now he's pushing the frontier further, focusing on building and scaling reliable AI agents through open-source research like DSPy and GEPA. Matei, this recognition is so well deserved. We're honored to build alongside you every day. awards.acm.org/about/2025-acm…

Decentralized RAG allows your database to benefit all LLM clients. On the other side, not all data sources are reliable. Managing source reliability on blockchain can avoid third-party manipulation. Introducing dRAG + Blockchain + Truth Discovery: arxiv.org/abs/2511.07577

𝗕𝗲𝗰𝗮𝘂𝘀𝗲 𝟵.𝟭𝟭>𝟵.𝟵 𝗮𝗻𝗱 𝗮 𝘁𝗿𝗶𝗮𝗻𝗴𝗹𝗲 𝗵𝗮𝘀 𝗳𝗼𝘂𝗿 𝘀𝗶𝗱𝗲𝘀, 𝘁𝗵𝗲𝗿𝗲𝗳𝗼𝗿𝗲 𝟭+𝟭=𝟮. LLMs and Language Agents can sometimes generate correct answers from blatantly incorrect reasoning, which is more often in complex tasks, and exacerbated by reinforcement learning (RL), the commonly believed silver bullet to complex reasoning in LLMs. This is due to a well-known phenomenon called reward hacking, where if the only training signal LLMs are getting from the training data exclusively regards the final result, then LLMs are incentivized to match the correct final output through whatever means possible on its training data, leading to inconsistent and ungeneralizable reasoning processes in RL's wake. With our intern Mengzhao Jia, we (@ignaciocases and myself, plus folks from @Meng_CS s lab at Notre Dame) explore a simple fix: can we use the LLM's own reasoning to provide some additional supervision signal for the reasoning process itself, so that besides the final result, the LLM is also encouraged to stay consistent in its reasoning during training? We design an algorithm to automatically create rubrics for LLM reasoning processes, and train the model to adhere to these rubrics alongside generating correct final answers during RL. The resulting model not only produces significantly more consistent reasoning, but also generalizes better on a wide range of complex reasoning tasks we benchmarked, even with just 10% of the training data. We hope this technique helps pave the way to more powerful and generalizable reasoning models for complex tasks. Read more in our preprint: arxiv.org/pdf/2510.14738

Thrilled to share that “Improving Large Language Models Function Calling and Interpretability via Guided-Structured Templates” paper has been accepted to EMNLP 2025 (Main Conference)!🎉 📄 Check it out on arXiv: arxiv.org/abs/2509.18076 project page: hygiadang.com/publication/em… 1/3

.@DomSoos from @WebSciDL and @oducs is presenting "Can LLMs Beat Humans on Discerning Human-written and LLM-generated Science News?" They explored whether LLMs can outperform humans for LLM-generated vs. human written news. 🔗doi: 10.1145/3720553.3746674 #LLM #NLP @fanchyna

@NoahZiems I've co-directed it for 6 months!!! lucyinstitute.nd.edu/centers-and-la…

@Meng_CS We have a foundation models lab?!

Job opportunity (postdoc at Notre Dame Foundation Models Lab): apply.interfolio.com/173333

✴️ Pleased to introduce our new paper yining610.github.io/dynamic-reward… - Rebalance multiobjectives during training through dynamic reward weighting - Build Pareto-dominant front over static baselines across online RL algorithms, datasets, and model families - Faster convergence rate 1/8

Come be my colleague! @ND_CSE at @NotreDame is hiring a tenure-track professor in computer vision! (And robotics and quantum.) More info here: careercenter.cra.org/job/university…

@lateinteraction @NoahZiems @MIT_CSAIL @DSPyOSS Congratulations to both of you! Glad to see the "interaction"/collaboration is getting strong and soon fruitful - and never too "late" :) @lateinteraction I am so excited too! Way to go, wonderful @NoahZiems ! Change the world with mind and hand!

@NoahZiems @MIT_CSAIL @Meng_CS @DSPyOSS Welcome Noah!! So great to have you as a founding member here of this new lab :D And I’m so excited to continue to collaborate with and learn more closely from Meng!

Quick update! This year I am on visit at @MIT_CSAIL working under the wonderful @lateinteraction while I am continuing to be advised by the wonderful @Meng_CS Right now my focus is to continue making Arbor a fantastic RL framework for optimizing @DSPyOSS programs

🔥 Only 15 days left! 🔥 The Open Polymer Challenge already has 9,800+ entrants and 38,000+ submissions. If you have not joined yet, let’s jump in these last few days to 🌍 accelerate polymer discovery with ML and go for 💰 $50,000 in prizes. 👉 LINK: kaggle.com/competitions/n…

New paper: Reflective Prompt Evolution Can Outperform GRPO. It's becoming clear that learning via natural-language reflection (aka prompt optimization) will long be a central learning paradigm for building AI systems. Great work by @LakshyAAAgrawal and team on GEPA and SIMBA.

@msbernst Congratulations!

Thank you to everyone for your energy and enthusiasm in joining this adventure with me so far!

🎓💰🔬 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… 📈

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???