Pinned
built an awesome thing with awesome people
A research preview of Operator, an agent that can use its own browser to perform tasks for you.
00:00
Reiichiro Nakano
1,611 posts
Reiichiro Nakano
@reiinakano
i like building awesome things with awesome people 🇵🇭 🇯🇵
- I am on an H-1B, in the process of getting my green card and relocating my family to the US. Me and many other colleagues in a similar situation have signed this letter. I do not know what will happen next, but I am confident we will be taken care of. the board should resign.This all happened after midnight, during a company wide holiday. Support was always near unanimous.
- Style transfer without a specific style image. I minimize only the content loss from neural style transfer and directly optimize brushstrokes via backpropagation. In this case, "style" is an intrinsic property defined entirely by the artistic medium.
00:00 - we switched out the underlying Operator model to o3 today. we think it's a step jump improvement from the previous 4o-based model. give it a try with some of your old prompts that failed! it was fun to work on this, the o-series paradigm makes everything so much better!Operator 🤝 OpenAI o3 Operator in ChatGPT has been updated with our latest reasoning model. operator.chatgpt.com
- New blog post: Neural Style Transfer with Adversarially Robust Classifiers I show that adversarial robustness makes neural style transfer work by default on a non-VGG architecture. Blog: reiinakano.com/2019/06/21/rob… Colab: colab.research.google.com/github/reiinak…
00:00 - At #NeurIPS2019, talking to a guy from DeepMind about how I can't reproduce their results because it's so resource intensive. "It's not too bad actually, we only use 16 GPUs, it's not like we use a cluster of TPUs or anything."
GIF - Some personal news: I moved to San Francisco and joined the OpenAI Fellows Program! Super excited to learn from all the amazing people here. It's a dream come true to be able to do machine learning full-time!
- Replying to @reiinakanoMy work is out on arXiv! Neural Painters: A learned differentiable constraint for generating brushstroke paintings: arxiv.org/abs/1904.08410 There are also @GoogleColab notebooks so you can reproduce all the experiments yourself: github.com/reiinakano/neu…
00:00 - Sharing my winter break project. I tried combining @hardmaru's world models and @yaroslav_ganin's SPIRAL to see if an agent can learn to paint inside its own dream. It can! These strokes are generated purely inside a world model, yet transfer seamlessly to a real paint program.
GIF - openai is nothing without its people
- New blog post: Solving simple probability problems with intermediate steps and an external calculator Link: reiinakano.com/2019/11/12/sol… A transformer is used to map questions to intermediate steps, while an external symbolic calculator evaluates intermediate expressions.
00:00 - Super excited to share that I joined Periodic Labs ~4 months ago! At OpenAI, I witnessed first-hand as RL broke benchmark after benchmark, and felt the big AI question increasingly shift from "how do we optimize?" → "what do we optimize?" To me, the answer is clear:Today, @ekindogus and I are excited to introduce @periodiclabs. Our goal is to create an AI scientist. Science works by conjecturing how the world might be, running experiments, and learning from the results. Intelligence is necessary, but not sufficient. New knowledge is
- I built a demo for arbitrary image stylization in the browser using @TensorFlow.js at reiinakano.github.io/arbitrary-imag…. Instead of having a single network per style like previous algorithms for fast style transfer, only a single model is used for all style images. more details in replies
