Xichen Pan

@xichen_pan

PhD Student @NYU_Courant, Researcher @AIatMeta; Multimodal Generation | Prev: @MSFTResearch, @AlibabaGroup, @sjtu1896; More at xichenpan.com

New York, USA

xichenpan.com

Joined August 2022

Pinned
Xichen Pan
@xichen_pan
Jun 16
Modern text-to-image models are increasingly powered by large pretrained LLMs. But there is a curious mismatch: the LLM typically encodes the prompt only once, while the evolving noisy latent states are handled entirely by a newly trained generative backbone. Can pretrained
Xichen Pan
@xichen_pan
Jun 18
Sharing an interesting fact about developing RepFusion. Last year, when I was working on MetaQuery, our goal was to let MLLMs generate images without training their understanding part. But I did not see much synergy between understanding and generation, and it did not outperform
Xichen Pan
@xichen_pan
Jun 16
Modern text-to-image models are increasingly powered by large pretrained LLMs. But there is a curious mismatch: the LLM typically encodes the prompt only once, while the evolving noisy latent states are handled entirely by a newly trained generative backbone. Can pretrained
Xichen Pan
@xichen_pan
Mar 20
There has been a lot of debate around the choice of denoising space. But it’s hard to get both semantics/diffusability and strong low-level reconstruction at the same time. REPA and VA-VAE are great explorations of adding semantics into the VAE space. After JiT came out, we
Han Lin
@hanlin_hl
Mar 18
🚀 Excited to share V-Co, a diffusion model that jointly denoises pixels and pretrained semantic features (e.g., DINO). We find a simple but effective recipe: 1️⃣ architecture matters a lot --> fully dual-stream JiT 2️⃣ CFG needs a better unconditional branch -->
Xichen Pan
@xichen_pan
Dec 16, 2025
(M)LLMs are effective at grounding and planning spatiotemporal layouts. Yet, they are mostly used as 1D conditional encoders in current generative and unified models. We explore an explicit interface to transfer these abilities to image editing and video generation.
Han Lin
@hanlin_hl
Dec 16, 2025
Multimodal LLMs (MLLMs) excel at reasoning, layout understanding, and planning—yet in diffusion-based generation, they are often reduced to simple multimodal encoders. What if MLLMs could reason directly in latent space and guide diffusion generation with fine-grained,
00:00
Xichen Pan
@xichen_pan
Jun 27, 2025
The code and instruction-tuning data for MetaQuery are now open-sourced! Code: github.com/facebookresear… Data: huggingface.co/collections/xc… Two months ago, we released MetaQuery, a minimal training recipe for SOTA unified understanding and generation models. We showed that tuning few
Xichen Pan
@xichen_pan
Apr 11, 2025
We find training unified multimodal understanding and generation models is so easy, you do not need to tune MLLMs at all. MLLM's knowledge/reasoning/in-context learning can be transferred from multimodal understanding (text output) to generation (pixel output) even it is FROZEN!