被卡脖子了

车尾灯都看不到了

最近国内大厂的科研热情达到了近几年的顶峰，虽然是被国外倒逼的，但好兴奋啊！

I wrote a blog post on why I decided to join OpenAI instead of academia. (after I went on the academic & industry job markets, and got offers from both.) This post (pt2 in a series) took a while 😅- hoping my experience helps others make life decisions! rowanzellers.com/blog/rowan-job…

🥂

I'm excited about Segment Anything released from FAIR today. It tackles an old problem (find objects in images) at large scale: trained on 11M images and 1B objects. This is a new Foundation Model for Computer Vision - it recognizes any object in any context.

Just made a demo space (unofficial ) of ImageBind for zero-shot image classification:) Hope it helps you figure out its embedding quality. Will add demos of other modalities soon. See huggingface.co/spaces/JustinL…

Happy to release ONE-PEACE, a general representation model towards unlimited modalities (vision, language, audio, vision-language, etc., for now). New SOTAs plus emergent zeroshot capabilties. Code to be released. abs: arxiv.org/abs/2305.11172 code: github.com/OFA-Sys/ONE-PE…

On the Chinese Valentines' day, we release the Multimodal (Vision-Language model), Qwen-VL, which is based on our previously released model Qwen-7B! This model can perform perform multi-image interleaved conversation (chat with images), visual grounding, text recognition, etc.

lucky to be a little part of the effort.

thanks for $STRK's airdrop!

这就投简历🫡

🍓

Thrilled to have two papers featured among paperdigest.org's most influential at top-tier AI conferences! 4DGS ranks 3rd out of 2719 CVPR'24 papers, and Vision Mamba ranks 2nd out of 2609 ICML'24 papers (1st if considering citation counts alone).

hero behind the scene