2022-03-16 14:26:39 RT @sharplm: @OpenAI has released a new version of its GPT-3 program that can not only generate text, but also edit and insert. This has hu… 2022-03-16 14:23:20 RT @nicksdjohnson: This new edit mode for GPT-3 is remarkable (h/t @jonty) https://t.co/l5QX0CsHXD 2022-03-16 04:39:58 RT @m0o0bav: Working with AI agents has to be intuitive and easy. So I and my teammates at OpenAI created new models that improve GPT-3 & 2022-03-15 20:22:08 RT @OpenAI: GPT-3 can now make changes to existing content, not just predict what comes next. Released in the API today: https://t.co/cDOWJ… 2022-03-15 19:29:46 RT @gdb: The OpenAI approach to making AI progress is to build working systems that can do something that was previously impossible. Succe… 2022-03-15 17:39:51 AI engineering (as opposed to AI science) is underrated, even today 2022-03-15 16:05:22 RT @woj_zaremba: "With Copilot, OpenAI is also offering a first peek at a world where AI predicts increasingly complex forms of thinking."… 2022-03-15 14:45:06 it is a privilege to work hard towards a great goal 2022-03-14 03:07:16 the ideas of deep learning are sublime in their simplicity and transcendent in their beauty 2022-03-13 20:19:35 there's an argument to be made that it's a moral duty to be happy 2022-03-12 18:04:07 in comfort we decay 2022-03-12 08:11:00 CAFIAC FIX 2022-01-17 08:11:00 CAFIAC FIX 2022-01-11 08:11:00 CAFIAC FIX 2021-12-27 08:20:00 CAFIAC FIX 2021-12-16 19:52:57 RT @reiinakano: I haven't tweeted much about ML in a ~year, but happy to finally share what I've been working on with amazing collaborators… 2021-12-16 17:53:49 problem solved https://t.co/xyf2CGy1q3 2021-12-16 17:52:21 A step towards automatically getting to the bottom of things https://t.co/3kMbxhIOrl 2021-12-16 17:51:29 RT @OpenAI: We trained a research version of GPT-3 that can search the web, synthesize information, and cite its sources to provide more ac… 2021-12-14 17:31:46 RT @OpenAI: Developers can now create a custom version of GPT-3 for their applications with a single command. Fine-tuning GPT-3 on your dat… 2021-12-01 00:16:39 RT @OpenAI: We're excited to announce the OpenAI Residency! This new program offers a pathway to a full-time role at OpenAI for researchers… 2021-11-18 23:25:49 RT @OpenAI: We introduced new safeguards to our API that make it possible to remove the waitlist. Now developers can sign up and explore GP… 2021-11-18 21:18:46 RT @sama: anyone who claims any certainty about what the impacts of AGI on the world are going to be is naive at best. we are heading into… 2021-11-18 20:59:07 RT @sama: We believe that the best way to carefully steward AGI into existence is to iteratively deploy increasingly powerful systems. A g… 2021-11-06 23:20:00 CAFIAC FIX 2021-11-01 19:20:00 CAFIAC FIX 2021-08-12 16:38:45 RT @woj_zaremba: Codex competition starts in 30 mins. Race the Internet to solve Python programming puzzles — together with Codex as both… 2021-08-12 03:01:19 RT @andrew_n_carr: The new @OpenAI Codex model is a pretty exciting piece of technology. Here I made a @Blender add-on and taught it how… 2021-08-11 20:01:02 RT @gdb: My personal favorite of our Codex examples — a program written in Python, which rewrites itself into Ruby, which rewrites itself i… 2021-08-10 22:59:56 RT @gdb: Giving Codex a first grade math test: https://t.co/PT7S0Pl7Sh 2021-08-10 20:33:55 RT @woj_zaremba: We are releasing an API to turn English into code. Make a computer game from scratch in a few minutes. https://t.co/fv8… 2021-08-10 19:05:11 RT @gdb: Top 500 finishers will receive OpenAI T-shirts! Looking forward to everyone getting to experience Codex for themselves, in a Pyth… 2021-08-10 18:49:09 RT @OpenAI: Announcing a new kind of programming competition. Race the Internet to solve Python programming puzzles — together with Codex… 2021-08-10 18:08:09 RT @OpenAI: OpenAI Codex has a qualitatively new capability — it can write code with sufficient accuracy that users can direct their comput… 2021-08-10 17:40:38 RT @sama: Today OpenAI launched Codex, which is an API that uses AI to write code from natural language. There's a demo video here: https… 2021-08-10 16:45:45 RT @woj_zaremba: Coding became fun to watch. See you in 20 mins on twitch! https://t.co/8zJE8S3b5n 2021-08-09 17:26:39 RT @OpenAI: See you tomorrow! https://t.co/Te3ZQsIO3v 2021-08-04 17:39:28 RT @OpenAI: Live demo of the next generation of OpenAI Codex, our AI that writes code, on Twitch at 10am PT, Tues Aug 10th. https://t.co/… 2021-07-28 17:04:05 Writing fast Cuda kernels just got a lot easier: https://t.co/c0yT4lhVNi 2021-07-22 17:30:31 Deep learning is a free lunch 2021-07-08 22:09:41 RT @janleike: There are a lot of exciting things in the Codex paper, but my favorite titbit is the misalignment evaluations by @BethMayBarn… 2021-07-08 02:32:09 RT @markchen90: Excited to share a paper detailing the capabilities and shortcomings of several early Codex models, whose descendants under… 2021-05-12 21:58:01 RT @prafdhar: 1/n Excited to release what @unixpickle and I have been working on for the past few months @OpenAI! We show diffusion models… 2021-05-04 17:38:23 RT @gdb: Welcome @HurdOnTheHill! https://t.co/rFzowbTmCJ 2021-05-03 01:54:45 the only thing more fixed than the supply of bitcoin is the supply of housing in California 2021-04-22 19:09:32 https://t.co/yRgSGpdcKg -- a detailed study of CO2 emission in large models from Google 2021-04-06 17:10:35 RT @gdb: New software engineering role open on my team. Will work closely with me and a small team of excellent folk (with a mix of skills… 2021-04-01 16:20:29 symbolic AI feature engineering is the future 2021-03-28 15:51:13 RT @karpathy: A new fun quick short story on AI: "Forward Pass" https://t.co/7ROcrHBqzI 2021-03-26 20:45:03 RT @OpenAI: A thought-provoking, interactive long read by @pamelamishkin, collaborating with GPT-3 to tell a very human story. https://t.co… 2021-03-25 19:44:27 RT @gdb: GPT-3 is currently generating an average of **4.5 billion words per day**: https://t.co/pqwhce8rD3 2021-03-16 18:37:55 RT @sama: AI is going to change a lot of things. The world is going to get phenomenally wealthy. I wrote about what I think will happen, a… 2021-03-08 19:03:15 RT @ItzhakFried: Voilà! Our multimodal concept neurons in human MTL (e.g.,Jennifer Aniston, Halle Berry neurons) https://t.co/t1l3joOeiI no… 2021-03-06 21:58:43 RT @MCHammer: Multimodal Neurons in Artificial Neural Networks #Hamm400aos https://t.co/mjqUHBGABR 2021-03-06 21:58:38 RT @MCHammer: #AI https://t.co/QzBp5bSHGu https://t.co/syKkdt6EE6 2021-03-04 20:41:45 RT @moyix: The latest generation of adversarial image attacks is, uh, somewhat simpler to carry out https://t.co/CBnA53lEcy https://t.co/h4… 2021-03-04 20:27:25 RT @OpenAI: We've found that our latest vision model, CLIP, contains neurons that connect images, drawings and text about related concepts.… 2021-03-04 00:38:54 RT @sama: Talking about the future of AI on Clubhouse tonight at 6 pm PT with @ilyasut @gdb and John Schulman: https://t.co/rNtdBEiZnE 2021-02-24 02:53:23 Vectors are symbols too 2021-02-22 21:41:54 RT @E0M: We're hiring platform & ML background not required. GPT-3 is just the beginning. Bringing safe AI… 2021-02-22 21:41:45 RT @npew: Come build the OpenAI API! https://t.co/PxrQuVIVrn 2021-02-20 01:30:54 The fundamental equation of deep learning: > 2021-02-19 04:21:09 Discrete optimization = Continuous optimization = 2021-02-17 19:38:02 RT @ryan_t_lowe: We're hiring again!! Our team works on both fundamental problems (how do we align models with what humans want?) and… 2021-02-17 03:35:46 RT @janleike: We're hiring research engineers for alignment work at @OpenAI! If you're excited about finetuning gpt3-sized language models… 2021-02-17 02:54:02 transformers: parallel computers in disguise 2021-02-10 21:17:59 RT @AndrewMayne: @ilyasut I wake up thinking about prompts. 2021-02-10 21:14:22 It helps to dream about ML 2021-02-10 20:41:20 RT @gdb: Agreed: https://t.co/1rZc1Dhgma https://t.co/vESf1DrBn4 2021-02-10 20:10:18 Luck favors the prepared mind https://t.co/4dXBI3gBpY 2021-02-10 19:44:20 Newton said, If others would think as hard as I did, then they would get similar results.'' https://t.co/FIkJtL0KqH 2021-02-10 19:35:03 Real progress in AI can only be achieved through a very intense work ethic 2021-02-10 14:02:35 Markets are the escape from Zero-sum 2021-02-09 23:14:05 RT @gdb: I don't have an academic degree. @ilyasut has a PhD (though dropped out of high school!). We do not select on the basis of academi… 2021-02-07 00:29:37 RT @lilianweng: If you are interested in developing innovative solutions for real-world machine learning problems and deploying cutting-edg… 2021-02-06 21:41:36 one of the most impactful future applications of LMs https://t.co/H9PLF72kvZ 2021-02-06 20:59:38 RT @manda_ngo: This makes me want to hug GPT-3 and believe in myself < 2021-02-05 22:40:11 RT @OpenAI: OpenAI research scientists use their combined skill and creativity to expand what’s possible with machine learning. Apply now t… 2021-02-04 03:19:34 Today, telling someone that they speak like a language model is an insult. But in the future, it'll be a compliment. 2021-02-03 23:33:22 RT @Hernandez_Danny: AI systems often use more direct experience than a human could get in a lifetime. Humans require less experience, beca… 2021-02-03 22:05:06 RT @OpenAI: We’re looking for research engineers to build and scale AI systems that achieve unprecedented levels of performance. Apply now… 2021-02-01 19:52:28 RT @OpenAI: Thinking about a career in AI research? The OpenAI Fellows program is looking for researchers from diverse backgrounds, with or… 2021-01-28 01:30:23 RT @DavidSacks: Censorship is a monster you cannot control. https://t.co/qfH6NXs4a8 2021-01-28 01:25:50 RT @sama: Most people will vigorously defend freedom of expression if and only if they agree with what you have to say. (This is certainly… 2021-01-26 01:21:39 RT @OpenAI: Scaling Kubernetes to 7,500 nodes: https://t.co/DSPU3zNufY 2021-01-20 20:55:32 RT @colinmegill: from july gpt-3 is still astonishing had the same outcomes and I want to tell you all something very serious and perso… 2021-01-14 02:40:24 RT @lilianweng: Our paper on training a single goal-conditioned policy 100% with asymmetric self-play to generalize to many unseen objects… 2021-01-14 00:57:55 amazed at the logo https://t.co/1OHLzCkJmm 2021-01-05 22:10:41 RT @prafdhar: Congrats to @AlecRad, Aditya Ramesh, @_jongwook_kim, @MikhailPavlov5, @gabeeegoooh, @scottgray76, @GretchenMarina, @SandhiniA… 2021-01-05 21:46:03 RT @_jongwook_kim: @seeingwithsound @ilyasut The smallest CLIP model is available at https://t.co/PS3Z0xT3bf. It’s larger than typical mode… 2021-01-05 21:10:33 An NN takes a list of category names, and outputs (in a zero-shot manner) a visual classifier. It beats RN50 on ImageNet zero-shot, while being far more robust to unusual images: https://t.co/dGdtFlbI7G https://t.co/tjYdMfFcfe 2021-01-05 20:23:56 Synthetic capybaras in different styles https://t.co/AXQ5mTIkuy https://t.co/gu6FCNwO6G 2021-01-05 19:43:31 CLIP: maps images to categories by taking class names as inputs DALL-E: text2im that works for a wide variety of sentences e.g., “a painting of an owl in a field at sunrise” below https://t.co/iSstNdbEqx 2020-12-11 19:12:57 Time flies when you’re having fun https://t.co/8TnI8PIs3G 2020-12-07 17:44:30 RT @OpenAI: Congratulations to the GPT-3 team for earning a Best Paper Award this morning at #NeurIPS2020! https://t.co/Gvc5brkZ5z 2020-12-05 02:22:56 RT @OpenAI: Join us at NeurIPS 2020 to learn more about OpenAI's research, including live demos of GPT-3: https://t.co/Gvc5brkZ5z 2020-11-30 23:18:38 @NandoDF Success is guaranteed 2020-11-30 16:18:04 Incredible! https://t.co/DtbsYjVebk 2020-11-12 22:19:27 RT @woj_zaremba: We are releasing a suite of robotics environments. Enjoy! https://t.co/X6HBC53E7L https://t.co/qrJ4A2vr6e 2020-11-03 20:56:04 “What I cannot create, I do not understand“ - Richard Feynman, making an early case for likelihood-based generative models 2020-11-01 16:21:12 The only thing I find more amazing than the rate of progress in AI is the rate in which we get accustomed to it 2020-10-25 22:20:26 The number of 2020 neurips submissions is greater than 0.5% of *all* scientific papers published in one year 2020-10-24 00:48:05 The fact that the brain is using backprop suggests it might be good to use in an NN too 2020-10-23 07:16:56 The fact that backprop is much better than all other NN methods suggests that the brain might be using backprop too 2020-10-21 23:34:28 It can be hard to tell where memorization ends and generalization begins 2020-10-19 22:01:11 wonder whether an airplane-shaped battery will have useful range 2020-10-18 20:09:57 SGD performs Bayesian Inference 2020-10-10 05:24:06 Matter tells spacetime how to curve Spacetime tells matter how to move -wheeler 2020-10-04 20:58:12 I wish there was as much interest in extremely cheap instant testing as there is in covid vaccine 2020-10-04 20:43:01 @lacker For example, modification to the architecture that make it more suitable to a problem class without changing fundamental properties (such as ease of optimization, information bottlenecks, etc) 2020-10-04 20:40:24 attention is all you need, as anonymous mathematical proof: https://t.co/uGTA7br5Bv 2020-10-01 22:15:43 The Bitter lesson does not say to not bother with methods research. It says to not bother with methods that are handcrafted datapoints in disguise. 2020-09-24 03:16:30 Neural networks are the muscles and bones of the spirit that is math 2020-09-23 20:06:07 I wish apple were to build a 16 inch iPad 2020-09-16 17:11:58 PPG, a new RL algorithm: More stable and more sample efficient than PPO on vision-based environments: https://t.co/XRRBp92Mbd 2020-09-14 04:37:37 1+2+3+4+….=-1/12 2020-09-10 03:46:19 Transformers can prove theorems! https://t.co/7Dq5161KfZ 2020-09-04 17:48:16 RT @OpenAI: We've used reinforcement learning from human feedback to train language models for summarization. The resulting models produce… 2020-09-04 02:31:00 the problem with academia is that its main contact with reality is through reviewer #2 2020-08-21 20:19:32 RT @woj_zaremba: GPT-3 is incredible in generating absurdity. This is a superpower in brainstorming. https://t.co/M402OJinIh 2020-08-15 15:16:55 RT @tusharkhattar_: Introducing Revtheo Browser Extension! Check it out in comparison with Apple's "word lookup feature". Revtheo, a conte… 2020-08-10 18:45:20 RT @BonJarber: Here are my initial results of integrating #GPT3 with Vector the robot this weekend. I don't have text to speech working ye… 2020-08-04 19:31:12 RT @gdb: OpenAI for learning a new language: https://t.co/xmevZJmfh9 2020-07-31 18:04:31 RT @gdb: We're hiring for the support team for our API: https://t.co/hIykpnQSWq Great way to interact with users across a wide-variety of… 2020-07-30 19:39:21 RT @gdb: An OpenAI-powered writing assistant, available on a smartphone near you: https://t.co/7MNKSQt4sV 2020-07-29 13:43:49 RT @gdb: One of our users is running a rigorous test of OpenAI's utility in copywriting: https://t.co/G6TYXWUNxN. You can sign up on the l… 2020-07-27 18:00:24 RT @gdb: Phil Brockman (my younger brother) just released his first Spotify album, in collaboration with @OpenAI MuseNet: https://t.co/28w7… 2020-07-27 16:07:35 RT @gdb: A curated list of how people have been applying OpenAI: https://t.co/mObY5H32oH 2020-07-25 22:05:27 RT @gdb: Great work @sh_reya &amp 2020-07-25 21:00:11 RT @gdb: Generate Keras models from a description. https://t.co/hhOjbJxa2b 2020-07-25 20:47:33 RT @gdb: Thank you from the team! Appreciate the support. https://t.co/WB8thDmci1 2020-07-25 20:44:41 You are too kind https://t.co/ECTnFsTJLO 2020-07-25 15:06:57 good meme https://t.co/N5XhQy9nva 2020-07-25 13:45:45 partly but not fully true https://t.co/NRQ9WzhZqy 2020-07-23 21:16:35 RT @gdb: We’ve received tens of thousands of applications for access to GPT-3 via our API. We want to serve everyone who wrote in, but it's… 2020-07-23 20:51:05 RT @OpenAI: Last month we released an API for developers to build on top of our latest technology: https://t.co/4BaTBWzRtQ. There’s been a… 2020-07-23 20:27:13 RT @Merzmensch: This is a love letter written by toaster: #gpt3 GPT-3 @OpenAI https://t.co/nL0d4VQGSG 2020-07-22 17:17:41 RT @sama: Hi Jerome! It's great to get feedback from someone with so much experience deploying AI at scale. We share your concern about bi… 2020-07-19 19:18:26 RT @gdb: Agreed. We're just getting started and have a *long* way to go. Incidentally, there's a great way to help us get there faster — j… 2020-07-19 18:45:47 RT @sama: The GPT-3 hype is way too much. It’s impressive (thanks for the nice compliments!) but it still has serious weaknesses and someti… 2020-07-17 16:28:53 RT @gdb: A prototype of a new class of education applications. https://t.co/vCuSOrzSMF 2020-07-17 15:47:56 RT @rewonfc: Thanks if you came to our ICML poster on Distribution Augmentation. The zoom discussion was way more fun/interesting than I ex… 2020-07-15 16:43:23 RT @sharifshameem: Here's a sentence describing what Google's home page should look and here's GPT-3 generating the code for it nearly perf… 2020-07-14 19:00:26 iGPT won an ICML outstanding paper honorable mention! https://t.co/kO8upfcPyI 2020-07-13 16:21:28 RT @sharifshameem: This is mind blowing. With GPT-3, I built a layout generator where you just describe any layout you want, and it genera… 2020-07-09 18:19:48 RT @OpenAI: Over the past 5 months our 3rd class of Scholars researched how GPT-2 represents grammar, measured the interpretability of mode… 2020-07-07 19:26:36 .@AlecRad, @markchen90 and I will be discussing some of our recent research (image GPT) and answering your questions, this Thursday, 7/9 from 10-11am PT. Please register here for the Zoom link to join us: https://t.co/8mhakSGjTC 2020-07-07 17:11:37 RT @gdb: OpenAI API for writing short stories in the style of a chosen author: https://t.co/zmOt8ZSuvy 2020-07-06 17:36:15 RT @gdb: Tweet written by OpenAI API: https://t.co/kSDB8VzNG2 2020-07-04 22:37:39 Programming is like herding cats. Except your cats are different, you have slightly different goals, and you're not actually herding. 2020-07-04 13:55:48 How i got my web browsing under control using minimal willpower: 1. Visit fb,twtr,etc only on an ipad, never on laptop/phone. 2. Set screen time with generous limits (3 hrs/day/site). 3. Reduce the screen time limit by 5 minutes each day. If it gets too hard, go back a step. 2020-07-03 17:23:40 RT @gdb: Applying the OpenAI API to generating creative ideas and content for marketing your work, by @_bramses: https://t.co/ZL1WtOGouZ 2020-06-27 21:18:57 Epilepsy causes seizures: the brain gets so overexcited that it malfunctions. One way to “cure” extreme epilepsy is to sever the connection between the hemispheres, making the brain less overconnected. Social media is the overconnection that causes epilepsy in society. 2020-06-26 22:59:03 @NandoDF @OpenAI @karpathy Thanks Nando! it was a large team effort by @nottombrown, @8enmann, Nick Ryder, Melanie Subbiah, @samsamoa, @alecrad and many others, and was led by @Dario_Amodei 2020-06-25 19:01:17 Hope is a choice 2020-06-24 14:33:34 RT @AndrewMayne: Can @OpenAI's API understand emojis and even use them to describe movies? Spoiler alert: Yes. https://t.co/s6eHuw5Rkp E… 2020-06-23 15:10:51 RT @ch402: I am deeply disappointed in @nytimes for its intent to doxx Scott Alexander. Scott is one of the most vibrant intellectuals I kn… 2020-06-19 17:38:16 RT @ch402: Software 1.0: Write an algorithm that has the right behavior. @karpathy’s Software 2.0: Optimize differentiable blob to have co… 2020-06-19 17:38:10 RT @karpathy: @ch402 https://t.co/5D75MvOs87 2020-06-18 01:30:33 RT @hardmaru: “Sequence modeling is a universal unsupervised learning algorithm… We deliberately chose to forgo hand coding any image spec… 2020-06-17 19:04:29 RT @sama: Congrats to the @MarkChe38147556, @AlecRad, @ilyasut, and the rest of the team that worked on this. The image completions are ama… 2020-06-17 19:04:19 RT @MarkChe38147556: Excited to share what I've been working on with @AlecRad, @rewonfc, @ilyasut and others! https://t.co/HsFvScvcri 2020-06-17 18:38:59 Transformers trained to predict pixels generate plausible completions of images and learn excellent unsupervised image representations! To compensate for their lack of 2d prior knowledge, they are more expensive to train. https://t.co/jSUl2697lv 2020-06-15 01:03:32 RT @danielgross: I fed federal tax law to OpenAI's model and asked for a summary of a section. Input on the left, output on the right. Pret… 2020-06-11 21:13:32 RT @gdb: Using the API is very simple. Just POST some text to us, and we'll stream back text from the model: https://t.co/8ahHQ2uBWj 2020-06-11 17:58:22 RT @OpenAI: More details on our blog: https://t.co/v1KmZVY4c8 2020-06-11 16:59:27 RT @nickwalton00: We've been using OpenAI's new API for the last several weeks and have been anxiously waiting till we could tell you abou… 2020-06-11 15:55:36 OpenAI API — making the smartest models easy to use in a safe way: https://t.co/wAR6cnfrox 2020-06-11 15:50:06 RT @woj_zaremba: We are releasing a powerful API to create chatbots, translate between languages, perform a semantic search, and summarize… 2020-06-11 15:50:05 RT @sama: A new API for OpenAI technology: https://t.co/Kirkqr0rZx It can do a lot today, and it will learn to do more. 2020-06-11 15:07:43 RT @OpenAI: We're releasing an API for accessing new AI models developed by OpenAI. You can "program" the API in natural language with just… 2020-06-09 16:13:21 RT @karlcobbe: My team at OpenAI is co-organizing two NeurIPS competitions this year using some of our most compelling RL environments: Pro… 2020-06-09 16:12:19 RT @OpenAI: We're co-organizing two NeurIPS 2020 competitions using Procgen Benchmark and MineRL. We rely heavily on these environments int… 2020-06-09 02:47:07 RT @OpenAI: We’ve been talking at OpenAI about what we can do to help Black equity. We’re committing $1M to this cause via direct donations… 2020-05-30 14:14:24 To improve our mood: https://t.co/srOFb4y5hZ 2020-05-26 23:39:38 We are each other's mirror neurons 2020-05-25 23:03:23 Physics is the process of learning to speak fluent universe — exurb1a 2020-05-24 13:29:23 RT @Guillegran: my plans 2020 https://t.co/vP99fIpajd 2020-05-17 22:32:28 the brain may as well be deterministic for all intents and purposes, yet it feels like we have free will. a possible reason is that our model of ourselves is not ourselves, so our decisions surprise our self model and feel like free will 2020-05-08 23:33:14 Had a lot of fun chatting with @lexfridman! https://t.co/gFZgFpBBd1 2020-05-06 12:35:43 I wish iOS/Android used a small fast NN to implement auto dark mode on all their apps 2020-05-05 16:06:54 A Moore’s law but for DL algorithms: https://t.co/ExjXzRWSFI 2020-05-04 16:50:37 RT @woj_zaremba: You give feedback every quarter and it’s a great time to positively affect your team and organization...so why does it fee… 2020-05-02 19:58:48 https://t.co/TlTaHrQEYT 2020-04-30 21:49:46 Cool TC article: https://t.co/HOLbGRMQ1T 2020-04-30 17:30:45 RT @prafdhar: In the early days, the model didnt know English (we didnt show it any lyrics) and so it used to just make words up. Led to so… 2020-04-30 17:30:43 RT @mcleavey: Hot Tub Christmas (with GPT-2 lyrics selected by @rewonfc ) is my own favorite -- even though the model can't figure out what… 2020-04-30 16:58:47 RT @rewonfc: One of my favorites from @OpenAI's jukebox: 'Lose Yourself' re-rendered by Kanye https://t.co/PlYKbaEjon 2020-04-30 16:49:54 RT @prafdhar: Thanks also to the amazing support from all the other teams at OpenAI on this release! Especially @apilipis for coordinating… 2020-04-30 16:45:59 RT @OriolVinyalsML: Very happy to see such a cool application of VQ-VAEs by our friends @OpenAI! Congrats! https://t.co/v01UvhnPRL 2020-04-30 16:40:27 RT @jluan: Give me an artist, genre, and lyrics (or not), and this neural network will generate you a song! It can even rap. This is one o… 2020-04-30 16:35:21 RT @prafdhar: Excited to finally release what we've been working on since June! Jukebox is a small step in making neural nets produce music… 2020-04-30 16:29:06 RT @rewonfc: awesome work @prafdhar @mcleavey, Heewoo! so cool! https://t.co/EJZ0Al3MrH 2020-04-30 16:16:22 RT @avdnoord: Nice use of VQ-VAEs for music generation! https://t.co/9Aeb2sIXoh 2020-04-30 16:14:57 RT @mcleavey: Thrilled to introduce Jukebox, our latest work generating music, now in the raw audio domain! With @prafdhar, Heewoo Jun, an… 2020-04-30 16:09:32 Large neural networks can be made to sing! :) https://t.co/EKLy9YaKJd 2020-04-30 16:05:54 RT @prafdhar: Wanna hear a neural net sing? Check out our latest work Jukebox! https://t.co/GVDLowMdQE https://t.co/1oKyG9H14t 2020-04-16 16:27:11 RT @OpenAI: We've contributed to a multi-stakeholder report by 58 co-authors at 30 organizations that describes 10 mechanisms to improve th… 2020-04-14 18:59:31 RT @ch402: Today @ludwigschubert @mpetrov @shancarter @nicklovescode @gabeeegoooh and I released the OpenAI Microscope: https://t.co/qxtVZ1… 2020-04-14 16:02:41 RT @OpenAI: Introducing OpenAI Microscope: a collection of visualizations of every layer and neuron in eight vision "model organisms" often… 2020-04-10 17:41:23 RT @Aravind7694: New paper - CURL: Contrastive Unsupervised Representations for RL! We use the simplest form of contrastive learning (insta… 2020-04-09 21:34:06 RT @sama: The supercomputing team is one of the most important teams at OpenAI, and they’re hiring. They build some of the largest compute… 2020-04-06 18:24:22 RT @woj_zaremba: At OpenAI, our Learned Optimizer team trains RL agents that tune hyperparameters, conduct experiments, and draw conclusion… 2020-04-05 12:35:10 RT @damocrat: Bill Gates to build factories to manufacture the 7 most promising vaccines, so they can all be tested in parallel. “A few bi… 2020-04-05 12:18:37 RT @michael_nielsen: Bill Gates is setting up factories to manufacture 7 leading vaccine candidates before we know which is best &amp 2020-04-04 19:29:45 https://t.co/POVXki5emY 2020-04-03 17:06:49 RT @woj_zaremba: Feeling a bit overwhelmed at work? Here are some things my team and I are doing to stay happy and healthy :) https://t.co/… 2020-03-20 18:39:54 the most stunning advance in graphics that i’ve seen in a long while: just infer the voxel grid with an MLP. Of course it’ll work! https://t.co/YrpOxrfXGT 2020-03-17 17:15:26 i like how it suggests that backprop may really be the best algorithm for training neural networks https://t.co/pCQJj8wFJ5 2020-03-14 17:53:41 RT @teamrework: VIDEO: Ilya Sutskever &amp 2020-03-10 16:24:31 Really compelling examples of how neural network features are compositional, in the sense of being literally composed of simpler features in ways that make sense! https://t.co/D66gaWJXOU 2020-03-07 17:10:54 this would remain true even if our emotions were much “smarter” — the stock market pits the intelligence of its participants against each other, so it would also become adversarial for the hypothetical smarter emotions while becoming more efficient overall https://t.co/E7kz3YLuXM 2020-03-06 17:17:47 The stock market is an adversarial example for human emotions 2020-03-02 20:39:07 Deep learning has fully solved the curse of dimensionality. It vanished like an RNN gradient! 2020-03-02 20:38:55 RT @OpenAI: OpenAI's multi-agent team is currently hiring research scientists! Interested in multiple, interacting agents, AI-generating al… 2020-02-23 20:46:45 The more exceptional one's life is, the more the world looks like a simulation 2020-02-19 20:39:32 Really cool nvidia GAN demo that turns anyone into a great artist: https://t.co/m9RBsZA6AP 2020-02-15 13:41:27 Found this site that generates custom song lyrics with a large NN: https://t.co/cTpHXgytrC 2020-02-13 21:16:49 RT @nicolehe: This was extremely fun: for @voguemagazine, I fine-tuned GPT-2 to generate interview questions for Billie Eilish!! The AI als… 2020-02-09 13:48:58 https://t.co/B9Ud0hET98 2020-02-07 16:04:57 https://t.co/8dyihEQqk5 — careful and expensive MCMC Bayesian inference over NN parameters is *worse* than point estimates or low temperature posteriors. Supports @carlesgelada and @jacobmbuckman’s view that Bayesian NNs are not meaningful probably because the prior is wrong. 2020-01-31 18:45:02 RT @lilianweng: Humans learn from curriculum since birth. We can learn complicated math problems because we have accumulated enough prior k… 2020-01-30 20:02:07 RT @woj_zaremba: OpenAI + PyTorch = https://t.co/lefeddMy52 2020-01-30 17:09:48 RT @PyTorch: Welcome OpenAI to the PyTorch community! https://t.co/YYr0vHzBDu 2020-01-30 17:08:48 RT @OpenAI: We're standardizing OpenAI's deep learning framework on PyTorch to increase our research productivity at scale on GPUs (and hav… 2020-01-17 04:18:37 https://t.co/44pZXB4ejS Cool theory paper presenting a problem that: - can be efficiently learned by SGD with a DenseNet with x^2 nonlin, - cannot be efficiently learned by any kernel method, including NTK. 2020-01-13 04:10:31 RT @eturner303: The funny thing about AIDungeon is due to all the GPT-2 pre-training on swaths of random web text, you can create an advent… 2020-01-09 16:49:28 RT @gdb: Watched "The Squire — AI Written Short Film". It's a live-action adaption of AI Dungeon gameplay. Super fun to see! https://t.co… 2020-01-06 18:13:51 Super excited to work with you, @jeffclune!!! https://t.co/w6jTBFQEew 2020-01-06 00:34:33 RT @gdb: I agree with this article that AI Dungeon 2 (the GPT-2 powered text-based game of infinite possibilities) is "one of the coolest v… 2020-01-03 19:06:11 Seconded! It'll be a truly great talk. https://t.co/7a1zVjku06 2019-12-23 03:10:12 This is amazing https://t.co/50rn41KpTo 2019-12-15 18:03:45 an amazing game https://t.co/XKxtnmk7VX 2019-12-14 09:14:52 @NandoDF Thanks Nando! 2019-12-14 01:03:16 RT @dota2updates: Dota 2 with Large Scale Deep Reinforcement Learning https://t.co/J45lBDxqQj Via @OpenAI #Dota2 https://t.co/yT0GLiTv56 2019-12-13 19:49:27 The DotA paper is out!!! https://t.co/IazVi6h04P 2019-12-09 17:46:06 Cool short video on the history of AI: https://t.co/WlOa5nEFaf 2001-01-01 01:01:01