Mu Cai @ Industry Job Market @MuCai7 Twitter Profile

Mu Cai @ Industry Job Market

@MuCai7

Ph.D. student @WisconsinCS, Multimodal Large Language Models Will graduate around 2025 May, looking for Research Scientist position around multimodal models.

144Posts 517Followers 405Following

Similar User

@xuefeng_du

@ZifengWang315

@Yang_Liuu

@jiefengchen1

@zihengh1

@yzeng58

@YiyouSun

@jzhao326

@blue75525366

@Changho_Shin_

@JiachenSun5

@Minghao__Yan

@zhmeishi

@jifan_zhang

@myhakureimu

Pinned

Mu Cai @ Industry Job Market

@MuCai7

28 May

Thanks for @_akhaliq 's sharing! (1/N) We propose M3: Matryoshka Multimodal Models, arxiv.org/abs/2405.17430 which (1) reduces the number of visual tokens significantly while maintaining as good performance as vanilla LMM (2) organizes visual tokens in a coarse-to-fine nested way.

AK

@_akhaliq

28 May

Matryoshka Multimodal Models Large Multimodal Models (LMMs) such as LLaVA have shown strong performance in visual-linguistic reasoning. These models first embed images into a fixed large number of visual tokens and then feed them into a Large Language Model (LLM).

Mu Cai @ Industry Job Market

@MuCai7

15 Nov

I am not in #EMNLP2024 but @bochengzou is in Florida! Go checkout vector graphics, a promising format that is completely different from pixels for visual representation. Thanks to LLMs, vector graphics are more powerful now! Go chat with @bochengzou if you are interested!

Mu Cai @ Industry Job Market

@MuCai7

19 Sep

VGBench is accepted to EMNLP main conference! Congratulations to the team @bochengzou @HyperStorm9682 @yong_jae_lee The first work for "Evaluating Large Language Models on Vector Graphics Understanding and Generation" as a comprehensive benchmark!

Mu Cai @ Industry Job Market

@MuCai7

7 Nov

Now TemporalBench is fully public! See how your video understanding model performs on TemporalBench before CVPR! 🤗 Dataset: huggingface.co/datasets/micro… 📎 Integrated to lmms-eval (systematic eval): github.com/EvolvingLMMs-L… (great work by @ChunyuanLi @zhang_yuanhan ) 📗 Our…

Mu Cai @ Industry Job Market

@MuCai7

15 Oct

1/N) Are current large multimodal models like #GPT4o really good at video understanding? 🚀 We are thrilled to introduce TemporalBench to examine temporal dynamics understanding for LMMs! Our TemporalBench reveals even the SOTA LMM #GPT4o achieves only 38.5, far from…

Mu Cai @ Industry Job Market Reposted

Mu Cai @ Industry Job Market

@MuCai7

4 Nov

Personal update: After 5.5 yrs at @MSFTResearch , I will join @williamandmary in 2025 to be an assistant professor. Welcome to apply for my PhD/interns. Interest: ML with foundation models, LLM understanding, and AI for social sciences. More information: jd92wang.notion.site/Professor-Jind…

Mu Cai @ Industry Job Market Reposted

Mu Cai @ Industry Job Market

@MuCai7

26 Oct

Fine-grained temporal understanding is fundamental for any video understanding model. Excited to see LLaVA-Video showing promising results on TemporalBench, @MuCai7! Yet, there remains a significant gap between the best model and human-level performance. The journey continues!

Jianwei Yang

@jw2yang4ai

16 Oct

🔥Check out our new LMM benchmark TemporalBench! Our world is temporal, dynamic and physical, which can be only captured in videos. To move forward, we do need LMMs to understand the fine-grained changes and motions to really benefit the downstream applications such as video…

Mu Cai @ Industry Job Market

@MuCai7

24 Oct

Great work on apply multi-granularity idea on image generation/manipulation! This share the same visual encoding design as our earlier work Matryoshka Multimodal Models (matryoshka-mm.github.io), where pooling is used to control visual granularity, leading to a multi…

Xihui Liu

@XihuiLiu

24 Oct

Introducing PUMA: a new MLLM for unified vision-language understanding and visual content generation at various granularities, from diverse text-to-image generation to precise image manipulation. rongyaofang.github.io/puma/ arxiv.org/abs/2410.13861 huggingface.co/papers/2410.13…