Adriel Saporta

AI is fast becoming a public utility, and foundational infrastructure on this scale should not be concentrated in the hands of a few powerful gatekeepers. Open models matter. I am beyond proud to be part of the team at @reflection_ai dedicated to building this more open future.

AI is fast becoming a public utility, and foundational infrastructure on this scale should not be concentrated in the hands of a few powerful gatekeepers. Open models matter. I am beyond proud to be part of the team at @reflection_ai dedicated to building this more open future.

@_ghorbani @reflection_ai welcome to the team @_ghorbani, pumped to have you on board!

Hi friends, after three incredible years at OpenAI I am excited to share that I am starting a new chapter at @reflection_ai, where I will be leading the Science of Scaling team. Our mission is to deepen the scientific understanding of large scale learning and to turn compute into intelligence as efficiently and predictably as possible.

Reflection will be at NeurIPS San Diego next week. Say hello to the team at the Reflection booth and come along to our panel on open AI ecosystems. You’ll hear from: @real_ioannis (Reflection) @joespeez (Meta, Llama & PyTorch) @natolambert (AI2, ATOM Project) @robertnishihara (Ray, Berkeley) @ying11231 (SGLang, former xAI) We’ll be discussing: What does it actually take to build open AI ecosystems? What can we learn from China's approach? What do transparency and sovereignty mean in practice? Where is frontier intelligence heading? Find the Luma link in the thread to mark it in your calendar.

amazing read from @Skiminok on why he joined @reflection_ai and on our mission to capture the open-weight frontier. I couldn't have put it better myself... join us! reflection.ai/careers

unbelievably proud of @rajpalleti314 and @rehaanahmad171, and grateful to have been able to advise such a brilliant and, most importantly, kind founding team. @askalphaxiv is going to reinvent how researchers discover and apply cutting-edge AI. this is only the beginning.

Today we're sharing the next phase of Reflection. We're building frontier open intelligence accessible to all. We've assembled an extraordinary AI team, built a frontier LLM training stack, and raised $2 billion. Why Open Intelligence Matters Technological and scientific progress is driven by values of openness and collaboration. The internet, Linux, and the protocols and standards that underpin modern computing are all open. This isn't a coincidence. Open software is what gets forked, customized, and embedded into systems worldwide. It's what universities teach, what startups build on, what enterprises deploy. Open science enables others to learn from the results, be inspired by them, interrogate them, and build upon them in order to push the frontier of human knowledge and scientific advancement. AI got to where it is today through scaling ideas (e.g. self-attention, next token prediction, reinforcement learning) that were shared and published openly. Now AI is becoming the technology layer that everything else runs on top of. The systems that accelerate scientific research, enhance education, optimize energy usage, supercharge medical diagnoses, and run supply chains will all be built on AI infrastructure. But the frontier is currently concentrated in closed labs. If this continues, a handful of entities will control the capital, compute, and talent required to build AI, creating a runaway dynamic that locks everyone else out. There's a narrow window to change this trajectory. We need to build open models so capable that they become the obvious choice for users and developers worldwide, ensuring the foundation of intelligence remains open and accessible rather than controlled by a few. What We've Built Over the last year, we've been preparing for this mission. We’ve assembled a team who have pioneered breakthroughs including PaLM, Gemini, AlphaGo, AlphaCode, AlphaProof, and contributed to ChatGPT and Character AI, among many others. We built something once thought possible only inside the world’s top labs: a large-scale LLM and reinforcement learning platform capable of training massive Mixture-of-Experts (MoEs) models at frontier scale. We saw the effectiveness of our approach first-hand when we applied it to the critical domain of autonomous coding. With this milestone unlocked, we're now bringing these methods to general agentic reasoning. We've raised significant capital and identified a scalable commercial model that aligns with our open intelligence strategy, ensuring we can continue building and releasing frontier models sustainably. We are now scaling up to build open models that bring together large-scale pretraining and advanced reinforcement learning from the ground up. Safety and Responsibility Open intelligence also changes how we think about safety. It enables the broader community to participate in safety research and discourse, rather than leaving critical decisions to a few closed labs. Transparency allows independent researchers to identify risks, develop mitigations, and hold systems accountable in ways that closed development cannot. But openness also requires confronting the challenges of capable models being widely accessible. We're investing in evaluations to assess capabilities and risks before release, security research to protect against misuse, and responsible deployment standards. We believe the answer to AI safety is not “security through obscurity” but rigorous science conducted in the open, where the global research community can contribute to solutions rather than a handful of companies making decisions behind closed doors. Join Us There is a window of opportunity today to build frontier open intelligence, but it is closing and this may be the last. If this mission resonates, join us.

💪💪💪

brilliant feature from @askalphaxiv—auto-generates a blog-style, easy-to-follow breakdown of any research paper! Here's what it created for my latest work, Symile: alphaxiv.org/overview/2411.…

alphaXiv now supports Claude 3.7 for understanding arXiv papers 🚀 While the input length is limiting for multi-paper questions, it delivers remarkably accurate analyses of pseudocode, often catching implementation subtleties Full integration with paper codebases coming soon 👀

@daisydwolf @zoejacoby @neal_katyal 🤩🤩🤩

Last week, I got to watch @zoejacoby crush it in her first argument before the Supreme Court. Even her former boss @neal_katyal wanted a photo with the star of the day. I couldn’t be a more proud friend 🥹

🛠️ We're providing everything you need: • full dataset in CSV format • pre-processed tensor data • code to generate the dataset from MIMIC-IV and MIMIC-CXR • best model checkpoint trained on Symile-MIMIC using the Symile objective Check it out! doi.org/10.13026/3vvj-…

🚨 Announcing Symile-MIMIC: a multimodal clinical dataset where each of the 11,622 samples includes an ECG and blood labs taken within 24 hrs of patient admission, along with a CXR taken 24-72 hrs post-admission. Perfect for exploring cross-modal relationships in medical data 🏥

@_rk_singhal very cool!! 🤩🤩

Got a diffusion model? What if there were a way to: - Get SOTA text-to-image prompt fidelity, with no extra training! - Steer continuous and discrete (e.g. text) diffusions - Beat larger models using less compute - Outperform fine-tuning - And keep your stats friends happy !?

alphaXiv has introduced communities, where members can stay connected with the latest research and discussions within a specific subfield. What tools and websites do you use to discover new research? Which features on them do you find most useful? Drop your answers below 👇

Alphaxiv is an awesome way to discuss ML papers -- often with the authors themselves. Here's an intro and demo by @rajpalleti314 at #neurips2024 .

Excited to have @akashnet and @togethercompute generously offering $20K in compute credits for top contributors in the newly formed Agents Community! Join here to stay up to date with the latest in agents research: alphaxiv.org/communities/ag…

love this so much 😍

Tired of maintaining your github research profile pages🤔? We are excited to introduce alphaxiv io, where you can now create your research profile with a click of a button 🚀 With just your name, we auto-generate your list of publications with images and a short editable bio!

🧠 Symile-M3 was specifically designed to test how well a model understands the higher-order relationships between all three modalities. Text and audio are both needed to predict the image—neither alone will work! 🤝 🤗 Start exploring: huggingface.co/datasets/arsap…

Symile-M3—a new massive multilingual, multimodal dataset of 33 million image 🖼️, audio 🎧, and text 📝 triples across 10 languages—is now LIVE on @huggingface! Perfect for all your multimodal modeling needs 💅