Dongping Chen

After playing with AI2Thor for half year, I was again and again realize its programmatical design with accurate gt provide the best playground for many projects. It is excited to see MolmoSpaces bring the Ai2Thor into next chapter. Can’t wait to try them!

Reward models make or break post-training for multimodal omni models (e.g., nano banana), yet there’s surprisingly little research on that‼️ We’re releasing MMRB2: new reward benchmark focusing on omni models, spanning T2I, editing, interleaved, and thinking with images 🧵1/n

@hq_fang @ManlingLi_ maybe?

I’m applying to PhD programs in robot learning this cycle and am actively looking for relevant opportunities in this space. My research focuses on developing generalist robotic manipulation policies that leverage strong priors, by jointly optimizing both the data and the models. If you know of opportunities or are open to chatting, please ping me!

Want to know what your LLM don’t know? This is how 👇 Preprint: arxiv.org/abs/2503.23361 Code: github.com/uscnlp-lime/SEA

📢📢New Preprint!📢📢 The Impact of Large Language Models in Academia: from Writing to Speaking arxiv.org/abs/2409.13686 TL;DR: Just the beginning, just a matter of time.

Toward Trustworthy Generative Foundation Models (GenFMs) 🚀 🎇After six months of hard work and thanks to the efforts of the entire team, our report on the trustworthiness of generative foundation models (GenFMs) has finally been released. 💡In this work, we: -Developed a standardized set of guidelines by analyzing global AI governance policies - Introduced TrustGen - a dynamic benchmarking platform for evaluating GenFM trustworthiness - Provided an in-depth discussion of the challenges and future directions for trustworthy GenFMs. 📜Check out our report: arxiv.org/pdf/2502.14296 Our toolkit is public at: github.com/TrustGen/Trust…

@MahtabBg @ch1m1m0ry0 @cydhsieh @ethnlshn @RanjayKrishna Congrats!

I'm exited to announce that our work (AURORA) got accepted into #CVPR2025🎉! Special thanks to my coauthors: @ch1m1m0ry0, @cydhsieh, @ethnlshn, @Dongping0612, Linda Shapiro and @RanjayKrishna, This work wouldn’t have been possible without them! See you all in Nashville 🎸!

Excited to share our #NeurIPS2024 spotlight: Acoustic Volume Rendering (AVR) for Neural Impulse Response Fields. AVR greatly improve the state-of-the-art in novel view spatial audio synthesis by introducing acoustic volume rendering. Listen with headphone for example below

Excited to present Task Me Anything at: 📄Poster @ Wed 4:30pm, East Exhibit Hall 🎙️Oral presentation @ Video LM workshop at Sat 1pm, East Meeting Room 13 Come or reach out if you want to chat about multi-modal models, synthetic data, benchmarks etc💜 🔗 To save events: tinyurl.com/tma-poster-1211 tinyurl.com/tma-talk-1214

Introducing AURORA 🌟: Our new training framework to enhance multimodal language models with Perception Tokens; a game-changer for tasks requiring deep visual reasoning like relative depth estimation and object counting. Let’s take a closer look at how it works.🧵[1/8]

🌟 Exciting Research in LLMs! 🌟 1/ Our latest paper is now available! Discover how we're pushing the boundaries of AI to prioritize both honesty and helpfulness in LLMs.

@HowieH36226 Nice work. GUI understanding is a very important and timely topic👏👏