https://t.co/U6BL9xVJX4

https://t.co/0qXWnHo9jt

Maybe this one is a little much https://t.co/RDG2qHfbwn

These three paragraphs from Kahneman in 2017 (pre-LLM) are something else - full of, as James says, "painful claims" that are grounded in a lifetime of research. https://t.co/DigUnucqMR

For these wondering, and as expected, Gemini 3 Deep Think solves the stack underflow bug that cost me a few days. The answer is a more decisive than Opus 4.5, the only other public model to solve it (even Gemini 3 Pro fails). It even points the exact location confidently. It takes forever though... I don't have harder tests for now, most my benchmarks are saturated and I'm super busy with SupGen stuff, so that's all I have to say about this one

Gemini 3 Deep Think - original music it simulated an orchestra, wow https://t.co/2jh4NO4S2f

My brain activity reveal https://t.co/j4z77Q6oXi

I tried the Kernel headset at the Brain and Body Foundation Model NeurIPS workshop https://t.co/ulTpd7mVU0

NeurIPS secret mug club https://t.co/U1GC0aKRnD

Someone told me yesterday they were going to the PI party and for a brief moment I was confused https://t.co/gpxfuDUWJ0

Happy Thanksgiving‼️‼️2016-2026 🎬10 Years later‼️ https://t.co/24iJ1Ixuwp

+1 more

Happy Thanksgiving‼️‼️2016-2026 🎬10 Years later‼️ https://t.co/24iJ1Ixuwp

+1 more

Best-selling Irish author Sally Rooney has told the high court that the proscription of Palestine Action under terror laws means she is unlikely to be able to publish new novels in the UK and may have to withdraw her current books from sale. The ban on Palestine Action as a terrorist organisation could also leave the Normal People author’s British publisher and the BBC - which has adapted two of Rooney’s works for TV - at risk of being accused of funding terrorism if they pay her royalties, she said. The effect on Rooney’s work was held up as an example of how the ban is impacting freedom of expression on the second day of the landmark judicial review into whether the government’s proscription of Palestine Action is lawful. Rooney has repeatedly affirmed her support for Palestine Action, which has targeted the sites and subsidiaries of Israeli arms company Elbit Systems in the UK amid Israel’s genocidal war on Gaza. In witness statements provided to the high court on 27 November, Rooney said Israel had committed genocide in Gaza – a view shared by a UN commission. She said Palestine Action's activity was from a “long and proud tradition of civil disobedience - the deliberate breaking of laws as an act of protest”. Rooney added: “I myself have publicly advocated the use of direct action, including property sabotage, in the cause of climate justice. It stands to reason that I should support the same range of tactics in the effort to prevent genocide.” Back in August, Rooney said she intended to use royalties from her work “to go on supporting Palestine Action”. Following this, Rooney said she was warned by the independent producer of her BBC dramatisations that any payment to her for those shows could be a breach of terrorism laws. In September, she was advised not to travel to the UK to accept the Sky Arts Award for Literature in recognition of her latest novel, Intermezzo. Rooney has also called out the “years-long imprisonment without trial” of Palestine Action activists and expressed fears for those currently on hunger strike. The final day of the judicial review is scheduled for Tuesday 2 December.

investing is easy https://t.co/Yi084CdC1d

investing is easy https://t.co/Yi084CdC1d

To celebrate the holiday season, we’re launching 25 Days of Agents - an advent calendar of exclusive deals from top AI companies to help you build your own agents. Every day until December 25th, we’ll unlock a new deal from partners including @Railway @Cloudflare @convex @vercel @n8n_io @huggingface @Lovable and more.

OMG there's a bookstore dedicated to technical books in Taipei! https://t.co/L0p6UBcats

Transformers v5's first release candidate is out 🔥 The biggest release of my life. It's been five years since the last major (v4). From 20 architectures to 400, 20k daily downloads to 3 million. The release is huge, w/ tokenization (no slow tokenizers!), modeling & processing. https://t.co/bFDbYw92Ur

With its v5 release, Transformers is going all in on #PyTorch. Transformers acts as a source of truth and foundation for modeling across the field; we've been working with the team to ensure good performance across the stack. We're excited to continue pushing for this in the future across training, inference, and deployment. - @matthew_d_white Executive Director, @PyTorch Foundation / GM of AI, @linuxfoundation

PyTorch Foundation is heading to NeurIPS 2025 with a full program of workshops, sessions, and community events, including our joint booth with Cloud Native Computing Foundation (CNCF) and the Open Source AI Reception with CNCF, Anyscale, Featherless AI, Hugging Face, and Unsloth AI. Connect with PyTorch maintainers and ecosystem contributors throughout the conference this week. What to expect: Workshop: Agentic Development at the Frontier December 2, 4:00–5:00 PM PT Featuring Joseph Spisak, Davide Testuggine, Aksel Joonas Reedi, Zach Wentz, Daniel Han, Sanyam Bhutani, and Aakanksha Choudhery 🔗 https://t.co/eFzdHy8Mes Workshop: On Device/Edge AI December 3, 12:00–1:30 PM PT Featuring Cagatay Bilgin, Matthias Lecher, Andrey Tovchigrechko, Varun Khare, Vijay Janapa Reddi, Karan Goel, and Mitesh Patel 🔗 https://t.co/YeDPq6Ry19 Evening Event: Open Source AI Reception December 3, 6:00–9:00 PM PT Hosted by PyTorch Foundation & CNCF with Anyscale, Featherless, Hugging Face, and Unsloth at Union Kitchen and Tap Gaslamp in San Diego 🔗 Register to secure your spot: https://t.co/CKO5cqb1ze Talk: Modular Agents and Orchestrated Coordination using Open Agentic Protocols December 4, 10:45–10:57 AM PT Featuring Matt White 🔗 https://t.co/tFtBmeiqqq Booth: Meet with Us in the Exhibit Sponsor Hall We’re proud to be a Diamond sponsor of NeurIPS 2025 alongside CNCF. Come visit us in the Exhibit Sponsor Hall A/B - Booth #723 #NeurIPS2025 #NeurIPS #OpenSource #OpenSourceAI #PyTorch

Training massive Mixture-of-Experts (MoE) models like DeepSeek-V3 and Llama 4-Scout efficiently is one of the challenges in modern AI. These models push GPUs, networks, and compilers to their limits. To tackle this, AMD and Meta’s PyTorch teams joined forces to tune TorchTitan and Primus-Turbo, AMD’s open source kernel library, for the new Instinct MI325X GPUs. Together, they reached near-ideal scaling across 1,024 GPUs, showing that efficiency and scale don’t have to be a trade-off. 📎 Read our latest blog: https://t.co/xcpdQpy8da #PyTorchFoundation #OpenSourceAI #TorchTitan #MoE

PyTorch Foundation is on site at NeurIPS 2025 with workshops, sessions, and community programming throughout the week. Stop by our joint booth with @CNCF and meet contributors at the Open Source AI Reception with @CloudNativeFdn, @anyscalecompute , @FeatherlessAI, @huggingface, and @UnslothAI Connect with PyTorch maintainers and ecosystem contributors throughout the conference this week: TODAY: Workshop: Agentic Development at the Frontier December 2, 4:00–5:00 PM PT Featuring @joespeez, Davide Testuggine, Aksel Joonas Reedi, @zkwentz, @danielhanchen, @bhutanisanyam1, and Aakanksha Choudhery 🔗 https://t.co/LWNRcnEHlc TOMORROW: Open Source AI Reception December 3, 6:00–9:00 PM Hosted by PyTorch Foundation & CNCF with Anyscale, Featherless, Hugging Face, and Unsloth at Union Kitchen and Tap Gaslamp in San Diego 🔗 Register to secure your spot: https://t.co/GAdpZAxw10 Workshop: On Device/Edge AI December 3, 12:00–1:30 PM PT Featuring Cagatay Bilgin, Matthias Lecher, Andrey Tovchigrechko, @AKAvkkhare, @profvjreddi, @krandiash, and Mitesh Patel 🔗  https://t.co/wElvm7nFk8 Talk: Modular Agents and Orchestrated Coordination using Open Agentic Protocols December 4, 10:45–10:57 AM PT Featuring @matthew_d_white 🔗 https://t.co/EK2QqceOAA Booth: Meet with Us in the Exhibit Sponsor Hall We’re proud to be a Diamond sponsor of NeurIPS 2025 alongside CNCF. Come visit us in the Exhibit Sponsor Hall A/B - Booth #723 #NeurIPS2025 #NeurIPS #OpenSource #OpenSourceAI #PyTorch

Our latest PyTorch Foundation Spotlight features @RedHat's Joseph Groenenboom and Stephen Watt on the importance of optionality, open collaboration, and strong governance in building healthy and scalable AI ecosystems. In this Spotlight filmed during PyTorch Conference 2025, they share how PyTorch serves as the nucleus of Red Hat’s generative AI stack. They discuss the role of @vllm_project in their inference work and why equitable contributions and clear governance standards are essential to a healthy open source community. 🔗 Watch the Spotlight: https://t.co/sQQVXsj3Qs #PyTorch #OpenSourceAI #OpenSource #AIInfrastructure

At #NeurIPS2025, @danielhanchen (@UnslothAI ), Davide Testuggine (@Meta), @joespeez (Meta), and @bhutanisanyam1 (Meta) led a focused discussion on how environments are shaping the next stage of agentic AI and reinforcement learning. Their session examined why environments are central to how agents act, learn, and are evaluated across simulated and real-world settings, covering scalable RL environment frameworks, safety and robustness benchmarks, high-performance simulators for heterogeneous hardware, and how environments integrate with trainers, inference engines, and post-training workflows that support alignment and deployment. 📸 More updates from #PyTorch at NeurIPS coming soon #AIInfrastructure #ReinforcementLearning

+1 more

Runhouse and Curavoice were recognized in the 2025 PyTorch Startup Showcase, with @DonnyGreenberg presenting for @runhouse_ and Shrey Modi presenting for @CuraVoice. Their work reflects the depth of innovation across AI infrastructure and applied AI in the PyTorch ecosystem. Thank you to our Startup Showcase MC @chappyasel (@_ai_collective) and our judges Irving Hsu (@MayfieldFund), Radhika Malik (Dell Technologies Capital), Kevin Crosby (GitHub), @denise_teng25 (@GradientVC), and @simontiu (@VertexVUS) for their thoughtful evaluation and support. 💡 Learn more about the 2025 Showcase winners and their work: https://t.co/dR120wWEV6 #PyTorchCon #PyTorch #OpenSourceAI #AIInfrastructure

Next Thursday’s Inside Helion: Live Q&A is hosted by Jason Ansel, @oguz_ulgen, @weifengpy, and Jongsok Choi from @Meta's PyTorch Compiler and Helion teams. It’s a chance to hear directly from the developers shaping Helion. Helion is a Python-embedded DSL that compiles to Triton, raises the level of abstraction for kernel authoring, and uses large-scale autotuning to surface the fastest implementations across hardware. This session will walk through the design philosophy, performance model, and what’s ahead on the roadmap. 🔗 https://t.co/aVnYgsdwis #PyTorch #AIInfrastructure #OpenSourceAI

In this new PyTorch Foundation Spotlight, Ankit Patel (@nvidia) shares how their teams build on PyTorch, from using it across projects to extending it into new areas like Tensor TLLM for inference optimization and Physics NeMo for physics neural networks. He also reflects on the energy and mix of newcomers and maintainers collaborating across the PyTorch community, and how this innovation is happening with and on PyTorch. 🎥 Watch the Spotlight: https://t.co/gWKpqVcHy4 #PyTorch #OpenSourceAI #AIInfrastructure

What is the future of intelligence? The answer could lie in the story of its evolution. Nice essay by @blaiseaguera published in @Nature. Essay: https://t.co/dh1Zu0QElI Some nice ideas in the piece: “Hunting is a prime example of this predictive modelling. A predator must predict actions that will get the prey into its stomach; the prey must predict the predator’s behaviour to stop that from happening. Starting in the 1970s, neuropsychologists and anthropologists began to realize that other intelligent entities are often the most important parts of the environment to model — because they are the ones modelling you back, whether with friendly or hostile intent. Increasingly intelligent predators put evolutionary pressure on their prey to become smarter, and vice versa.” “Humans did not invent computation any more than they did electric current or optical lenses. We merely re-discovered a phenomenon nature had already exploited, developed mathematical theories to understand it better and worked out how to engineer it on a different substrate.”

Proud to see many Canadians at #NIPS2016 🇨🇦 https://t.co/8i41Hud47K

This is my 10th time attending #NeurIPS conference. The first time was in 2016 ✈️ https://t.co/oYugj8QQ5v

Sakana AIって最近資金調達したけど、どんな事業開発してるの？という方に是非見て頂きたいです🐟 金融（Finance）と防衛（Defense）の領域に注力していて、社内でとてもエキサイティングなプロジェクトが沢山あり、すごい成果も出てきてます 皆さんにSakana AIで働いてみたい！と思っていただけると嬉しいです🚀 https://t.co/szllV6tQk6

Fukushima's video (1986) shows a CNN that recognises handwritten digits [3], three years before LeCun's video (1989). CNN timeline taken from [5]: ★ 1969: Kunihiko Fukushima published rectified linear units or ReLUs [1] which are now extensively used in CNNs. ★ 1979: Fukushima published the basic CNN architecture with convolution layers and downsampling layers [2]. He called it neocognitron. It was trained by unsupervised learning rules. Compute was 100 times more expensive than in 1989, and a billion times more expensive than today. ★ 1986: Fukushima's video on recognising hand-written digits [3]. ★ 1988: Wei Zhang et al had the first "modern" 2-dimensional CNN trained by backpropagation, and also applied it to character recognition [4]. Compute was about 10 million times more expensive than today. ★ 1989-: later work by others [5]. REFERENCES (more in [5]) [1] K. Fukushima (1969). Visual feature extraction by a multilayered network of analog threshold elements. IEEE Transactions on Systems Science and Cybernetics. 5 (4): 322-333. This work introduced rectified linear units or ReLUs, now widely used in CNNs and other neural nets. [2] K. Fukushima (1979). Neural network model for a mechanism of pattern recognition unaffected by shift in position—Neocognitron. Trans. IECE, vol. J62-A, no. 10, pp. 658-665, 1979. The first deep convolutional neural network architecture, with alternating convolutional layers and downsampling layers. In Japanese. English version: 1980. [3] Movie produced by K. Fukushima, S. Miyake and T. Ito (NHK Science and Technical Research Laboratories), in 1986. YouTube: https://t.co/MUyH81L5wD [4] W. Zhang, J. Tanida, K. Itoh, Y. Ichioka. Shift-invariant pattern recognition neural network and its optical architecture. Proc. Annual Conference of the Japan Society of Applied Physics, 1988. First "modern" backpropagation-trained 2-dimensional CNN, applied to character recognition. [5] J. Schmidhuber (AI Blog, 2025). Who invented convolutional neural networks? https://t.co/chfcmk253b