Pan Lu Jeff Dean (@🏡) Google DeepMind

Congrats, Jeff Dean (@🏡) Google DeepMind! Gemini 1.5 Pro has shown substantial improvements from Feb to May, scoring 63.9% on our #MathVista (mathvista.github.io), outperforming humans and GPT-4o, which was out 4 days ago!🚀

AI Progress has never been this rapid and impressive!🌟

Thrilled to announce that I am joining UCLA CS Department as an Assistant Professor this Fall! 🐻

Many thanks to my incredible advisors, mentors, family and friends for the encouragement and support. ❤️Looking forward to this exciting new chapter and all the opportunities ahead! 🤖🤖🤖

Such a fun coincident on picking the same name.

Before scaling up, we called it CM3leon (pronounced as chameleon, with a twist to older cm3 paper) last year, in the paper 'Scaling Autoregressive Multi-Modal Models: Pretraining and Instruction Tuning' (arxiv.org/abs/2309.02591).

QRData is accepted at #ACL2024 Findings 🎉 Many thanks to our collaborators Zirui Wu Xueqing Wu Pan Lu Kai-Wei Chang and Yansong!

🤔Naming things is hard!!

🦎 #Meta 's new work shares the same name as our NeurIPS 2023 paper from one year ago: Chameleon: Compositional Reasoning with LLMs.

Coincidence or great minds thinking alike? 😈 Dive into our work here:
chameleon-llm.github.io
arxiv.org/abs/2304.09842

📢In the age of Generative AI, when people talk about how AI can transform various fields, the question arises: to what extent can AI actually assist clinicians, pathologists, and researchers in medicine, biology, and pathology? I was fortunate to be invited by spatial omics

Beats humans on Mathvista

This week at #ICLR2024 , researchers present MathVista, an open-source benchmark for measuring foundation models’ mathematical reasoning capabilities in multimodal scenarios. Learn about the work now from coauthor Michel Galley in the “Abstracts” podcast. msft.it/6011YpuIX

A fitting wrap-up to my ICLR 2024 ✨

OpenAI GPT-4o benchmarked and made significant advancements on our #MathVista dataset, achieving a whopping score of 63.8%!

Pan Lu Kai-Wei Chang uclanlp

openai.com/index/hello-gp…

First Falcon 2 released! 🚀 Technology Innovation Institute released Falcon 2 11B, the first model of the Falcon 2 family. Falcon 2 11B is a newly trained dense decoder model on 5.5 trillion tokens with a Vision Language Model coming soon! 🖼️

Model: huggingface.co/tiiuae/falcon-…

TL;DR:
🔢 Base LLM with 11B

Congrats!

Lumina-T2X: Transforming Text into Any Modality, Resolution, and Duration via Flow-based Large Diffusion Transformers

repo: github.com/Alpha-VLLM/Lum…
abs: arxiv.org/abs/2405.05945

LLaVA-NeXT is Expanded: Support Larger Models and Video Tasks

-🖼️Stronger LLMs Supercharge Multimodal Capabilities in the Wild
🗞️Blog 1: llava-vl.github.io/blog/2024-05-1…

-📽️A Strong Zero-shot Video Understanding Model
🗞️Blog 2: llava-vl.github.io/blog/2024-04-3…

-🌠Code: github.com/LLaVA-VL/LLaVA…