Xiaochuang Han
@XiaochuangHan
Followers
692
Following
1K
Media
16
Statuses
103
PhD student at the University of Washington
Seattle, WA
Joined February 2016
RT @shangbinfeng: 👀 How to find more difficult/novel/salient evaluation data?.✨ Let the data generators find it for you!. Introducing Data….
0
15
0
RT @orevaahia: 🎉 We’re excited to introduce BLAB: Brutally Long Audio Bench, the first benchmark for evaluating long-form reasoning in audi….
0
47
0
Check out our work led by @Cumquaaa on a hybrid autoregressive-diffusion architecture for image generation -- it flexibly balances the number of autoregressive and diffusion layers for optimal generation quality and inference speed! Autoregressive vs. diffusion -- you don't have.
🚀 Training an image generation model and picking sides between autoregressive (AR) and diffusion? Why not both? Check out MADFormer with half of the model layers for AR and half for diffusion. AR gives a fast guess for the next patch prediction while diffusion helps refine the
0
6
21
RT @WeijiaShi2: Can data owners & LM developers collaborate to build a strong shared model while each retaining data control?. Introducing….
0
87
0
RT @ZEYULIU10: LLMs trained to memorize new facts can’t use those facts well.🤔. We apply a hypernetwork to ✏️edit✏️ the gradients for fact….
0
66
0
RT @lpjiang97: I'm excited to share our latest work — "Data Heterogeneity Limits the Scaling Effect of Pretraining in Neural Data Transform….
biorxiv.org
A key challenge in analyzing neuroscience datasets is the profound variability they exhibit across sessions, animals, and data modalities–i.e., heterogeneity. Several recent studies have demonstrated...
0
13
0
RT @liujc1998: Today we're unveiling OLMoTrace, a tool that enables everyone to understand the outputs of LLMs by connecting to their train….
0
48
0
RT @michiyasunaga: 📢 Introducing Multimodal RewardBench:. A holistic, human-annotated benchmark for evaluating VLM reward models or judges….
0
37
0
RT @WeijiaShi2: Introducing 𝐋𝐥𝐚𝐦𝐚𝐅𝐮𝐬𝐢𝐨𝐧: empowering Llama 🦙 with diffusion 🎨 to understand and generate text and images in arbitrary sequen….
0
180
0
RT @alisawuffles: very excited to attend the first @COLM_conf!🤩 I will be presenting proxy-tuning as a spotlight talk🔦 on Wed @ 10am. I wo….
0
10
0
RT @vidhisha_b: On some personal news: I joined MSR AI Frontiers a few months ago and am very excited to share my first work with this amaz….
arxiv.org
Rigorous and reproducible evaluation is critical for assessing the state of the art and for guiding scientific advances in Artificial Intelligence. Evaluation is challenging in practice due to...
0
11
0
RT @gh_marjan: Can we train an LM on raw JPEG bytes and generate images with that? Yes we can. Check out JPEG-LM (,….
arxiv.org
Recent work in image and video generation has been adopting the autoregressive LLM architecture due to its generality and potentially easy integration into multi-modal systems. The crux of...
0
6
0
RT @PangWeiKoh: Check out JPEG-LM, a fun idea led by @XiaochuangHan -- we generate images simply by training an LM on raw JPEG bytes and sh….
0
4
0
Check out our paper ( for more details! Apart from the experiments and analyses of JPEG-LM, we also include proof-of-concept video generation experiments using canonical video codecs (AVC/H.264). Joint work with @gh_marjan , @PangWeiKoh , and Yulia.
arxiv.org
Recent work in image and video generation has been adopting the autoregressive LLM architecture due to its generality and potentially easy integration into multi-modal systems. The crux of...
0
1
8
The style of this work follows ImageGPT ( but the use of compression algorithms like JPEG greatly reduces the length constraint (e.g., by 40x). Compared to vector quantization methods, JPEG-LM is extremely simple (without any sophisticated tokenizer
1
0
2
👽Have you ever accidentally opened a .jpeg file with a text editor (or a hex editor)?. Your language model can learn from these seemingly gibberish bytes and generate images with them!. Introducing *JPEG-LM* - an image generator that uses exactly the same architecture as LLMs
4
39
157
RT @violet_zct: Introducing *Transfusion* - a unified approach for training models that can generate both text and images. .
0
209
0
RT @liliyu_lili: 🚀 Excited to share our latest work: Transfusion! A new multi-modal generative training combining language modeling and ima….
0
17
0