Jerry Li

We built an AI writer arena: 5 AI writers compete, evolve plots, and get shaped by AI readers’ feedback. It could be an IP factory for short-form drama 🎬 Built in a weekend by @JerryLiJiaming, @ywang485, and me → ipfactory.cc What we learned 👇

@drpeepee Thank you for the tips! Will keep them in mind I'm pretty excited if we can have real time models like genie 3 that helps us plan and test out different camera moves quickly.

Hey! Animator here - this model is pretty damn performant! The volumes are like 95% intact and the asset tracking is pretty immaculate One thing that could be helpful is plotting out the camera moves so you pick strategic moments where you have pans, dollies and tracking shots. Adding sound effects and music can go along way and can help bridge the continuity from shot to shot Sick model!

A year ago, I left DeepMind to chase a crazy dream with Spellbrush: to bring anime characters to life. Last week, my colleague and I put our in-house AI models to the test. We made this short anime from scratch in 48 hours. 1/3

@SwayStar123 sick! which model did you use?

@JerryLiJiaming Pretty decent but alot of room for improvement! I also tried making some stuff using ai a while ago x.com/SwayStar123/st…

@Abe_Froman_SKC Thanks! Will strive to make it better each year.

@JerryLiJiaming Cool tech, lame video.

@nikolodis @bubbleboi the video model used is already available on midjourney! updates.midjourney.com/introducing-ou… The text-to-speech model is in-house. We may make it an API service if we can get it to production quality.

@JerryLiJiaming @bubbleboi when will this be available for testing ?

We are looking for: 👉 LLM researchers jobs.ashbyhq.com/spellbrush/b57… 👉 3D Motion researchers jobs.ashbyhq.com/spellbrush/c77… 👉 Unity developers jobs.ashbyhq.com/spellbrush/977… If you know someone who would love to push the boundaries of anime and AI, share this with them—or join us yourself 3/3

Growing up, I loved anime and games as an art form. To finally cross to the other side—to build the worlds I once only imagined—feels surreal. We’re working on a 3D first-person adventure game, where the core of the experience is an AI character you can truly connect with. 2/3

【お知らせ】Niji Videoがついにリリース！✨ （音声あり🎶） リリースを記念して、テーマソングをご用意しました！ ぜひ新機能とあわせてお楽しみください。 使い方の紹介はリプライ欄から👇

We watched #GoogleIO with Project Astra. 👀

Relari just launched on @ycombinator's Launch YC! Congratulations to Yi's team! Relari - Testing and simulation stack for GenAI systems. Check them out: ycombinator.com/launches/KdO-r…

@gordic_aleksa @DeepMind Good luck! Glad that you are pursuing your passion.

BIG LIFE ANNOUNCEMENT: I'm leaving @DeepMind to start my own company. I'm 28 now. This is a start of a new life chapter. I'm both happy and sad. ❤️ I'm happy because I've been planning on starting my own company ever since I graduated from college back in 2017. 1/ (MEGA 🧵)

Mixed Integer Programming is an NP-hard optimisation problem arising in planning, logistics, resource allocation, etc. Presenting a solver with neural heuristics that learns to adapt to the problem domain, outperforming SCIP on Google-scale MIPs: dpmd.ai/13349 (1/)

Crypkoがスマートフォン向けアプリケーション『進化する少女型情報体 MEMES/ミームズ』（@prj_MEMES）のコアテクノロジーとして提供されます！ wfs.games/memes/ #Crypko #MEMES #ミームズ

@huaji0353 Glad that the dataset was useful! How many more images did you annotate from pixie?

RL Unplugged🔌: Offline RL benchmark that comes with both data and implementations of existing offline RL agents. It makes it easy to enter RL research and allows reproducibility.More agents in Acme will come soon. Github: git.io/JJUhd arxiv: arxiv.org/abs/2006.13888

Check out the new blog post by @notwaldorf! Extending previous work, she shows you can personalize a pretrained MuiscVAE model by training a tiny VAE in the latent space. Efficiently trained (in the browser!), and on just a couple of data points. magenta.tensorflow.org/midi-me

紙と厚紙と鉛筆などを使って工作しました。 #花火 #夏祭り #夫婦 #認知症

←なんで怒ってるかわからない俺 →その態度がさらにムカつく嫁

@alanyttian 楼顶的美术馆？赞