🤝 Can LLM agents really understand us?
We introduce UserBench: a user-centric gym environment for benchmarking how well agents align with nuanced human intent, not just follow commands.
📄 https://t.co/1dnzD3il0t
💻 https://t.co/hZuI2ZQTKE
Can LLMs make rational decisions like human experts?
📖Introducing DecisionFlow: Advancing Large Language Model as Principled Decision Maker
We introduce a novel framework that constructs a semantically grounded decision space to evaluate trade-offs in hard decision-making scenarios transparently.
📑Paper: https://t.co/ItbbRHbjCL
💻Code: https://t.co/74HduQTfMY
🧵👇
🚀 Can we cast reward modeling as a reasoning task?
📖 Introducing our new paper:
RM-R1: Reward Modeling as Reasoning
📑 Paper: https://t.co/VxuZ8JuhUJ
💻 Code: https://t.co/R583Hib26g
Inspired by recent advances of long chain-of-thought (CoT) on reasoning-intensive tasks, we hypothesize and validate that integrating reasoning capabilities into reward modeling significantly enhances RM's interpretability and performance. RM-R1 achieves state-of-the-art or near state-of-the-art performance of generative RMs on RewardBench, RM-Bench and RMB.
🧵👇
📢📈 I’m on the 2025 faculty job market!
I've been incredibly grateful to work with inspiring advisors, mentors & peers.
💡My research, AI4Scientists🔬, accelerates & democratizes the research lifecycle by:
1️⃣ Few-shot scientific knowledge acquisition
2️⃣ Domain-aware scientific LLM reasoning
3️⃣ Explainable scientific knowledge dissemination
I'm wrapping up my @uiuc_nlp Ph.D. (2025).
If you’re interested in those topics, I’d love to discuss further or have a chat!
I am at #EMNLP2024!
I will present our work "Why Does New Knowledge Create Messy Ripple Effects in LLMs? " on Web 10:30am.
Thanks to all the collaborators @hengjinlp@zhangzxUIUC@Glaciohound@ManlingLi_
Looking forward to have a chat!
Paper Link: https://t.co/AZz83MIp4J
I am at #EMNLP2024!
I will present our work "Why Does New Knowledge Create Messy Ripple Effects in LLMs? " on Web 10:30am.
Thanks to all the collaborators @hengjinlp@zhangzxUIUC@Glaciohound@ManlingLi_
Looking forward to have a chat!
Paper Link: https://t.co/AZz83MIp4J
In our work, we investigate the ripple effect problem in Knowledge Editing. We propose an indicator GradSim to estimate the performance of Ripple Effect post-editing.