Your curated collection of saved posts and media
Jensen Huang: "If that $500,000 engineer did not consume at least $250,000 worth of tokens, I am going to be deeply alarmed. This is no different than a chip designer who says 'I'm just going to use paper and pencil. I don't think I'm going to need any CAD tools.'" https://t.co/CcoYHRgrpy
Proteins can now talk. Introducing BioReason-Pro, the first reasoning model for protein function. A threadπ§΅
PyTorch 2.10 is now optimized for @Intel Core Ultra Series 3 processors to bring high-performance AI to the PC and edge. This release leverages the new Xe3 architecture and Arc B-series GPUs to deliver up to 120 XMX TOPs. Native TorchAO integration enables seamless int4 weight-only quantization, allowing a Llama 3.2 3B Instruct model to run locally with a first token latency of 242.29 ms and a 2+ token latency of 27.24 ms. For edge scenarios, developers can achieve 1.4x to 1.7x faster training for vision models via Anomalib, with WinClip seeing gains up to 2.5x compared to previous generations. Read our latest blog from the Intel PyTorch and Client AI SW teams for the full technical deep dive and benchmarks: https://t.co/pEGpmwtROi #PyTorch #Intel #AIPC #EdgeAI #OpenSourceAI
Nvidia has nearly doubled over the last two years. (12x over last five). But itβs fallen since GPT-5 came out. People are no longer buying the hype. https://t.co/uSq8hwYT3M
Jesus CHRIST https://t.co/omIIejTdxt
An Ohio-based company that initially balked at granting a mom's request to work from home during a high-risk pregnancy has been found liable for the newborn's death and ordered to pay $22.5 million in damages. https://t.co/OTnjOtalXc
Jesus CHRIST https://t.co/omIIejTdxt
Thread on VJEPA 2.1π€ This DEFINITELY flew under the radar: just a few days ago, @AIatMeta released V-JEPA 2.1, taking a massive step toward closing the gap between image and video domains. For a long time, image backbones were the only option for solving dense vision tasks. This model disagrees, showing that universal spatial understanding also emerges from large-scale video models!π₯
4/ The nonsense beats coherent prose in EVERY configuration. 18 out of 18. GPT-5, 5.1, 5.2, 5.3, 5.4. With reasoning, without, maximum reasoning. The nonsense wins every time. https://t.co/WxfV3kW96f
I've got a new profile of Alex Boresβrunning for Congress in Manhattan and the prime target of the AI-industry-backed Super PAC Leading the Futureβout today for @POLITICOMagazine. Some reporting nuggetsβ¦ https://t.co/1TOf4C1fZe
Range. https://t.co/70PMz76oXX
Range. https://t.co/TRAmdOrRTX
Range. https://t.co/70PMz76oXX
Loc3R-VLM Language-based Localization and 3D Reasoning with Vision-Language Models paper: https://t.co/grVjjllsho https://t.co/438E5UZHIB
even the trigger warning for that AI-generated novel is obvious slop lol https://t.co/mCoCz6vawE
LlamaParse now has an official Agent Skill you can use across 40+ agents. With built-in instructions for parsing complex documents, including different formats, tables, charts, and images, your agents gain access to deeper document understanding, not just raw text extraction. π Watch the demo π Read the docs: https://t.co/gRitu8byJe π Get started with LlamaCloud: https://t.co/YzFSMiUj0k
SAMA Factorized Semantic Anchoring and Motion Alignment for Instruction-Guided Video Editing paper: https://t.co/mhaPHEKtc5 https://t.co/OyXfM4NhSQ
FASTER Rethinking Real-Time Flow VLAs paper: https://t.co/1BtEOWx73Q https://t.co/EG8Gguj4vG
What representation enables open-world robot manipulation from generated videos? Introducing Dream2Flow, our recent work that bridges video generation and robot control with 3D object flow. https://t.co/vpRxoNBVF3 @Stanford #ICRA2026 1/N
Behind the scenes at @NVIDIAGTC: @clattner_llvm and @LambdaAPI's Sam Khosroshahi on how Modular and Lambda are shaping the future of AI infra. Full conversation dropping soon on Lambda's YouTube: https://t.co/K2eRHFLNvy https://t.co/CMipMsC0R5
Its good to be back at @huggingface spaces of the week :) https://t.co/vtoaykrbq6
Its good to be back at @huggingface spaces of the week :) https://t.co/vtoaykrbq6
john cassavetes in mikey and nicky (1976) https://t.co/3QUtMnNZns
AI agents are stepping into commerce with real economic power. They now have their own USDC wallets and make automatic micropayments on their own. @corey__cooper (Senior Manager, Developer Relations & Ecosystem Marketing @circle) Γ @czech_pawel (LabLab AI Co-Founder & @Surgexyz_ CEO) sit down for the Agent-to-Agent commerce talk everyoneβs been waiting for. Join us live this Wednesday (March 25) at 10AM PST.
Lil Elon chibified by @Grok Imagine https://t.co/yw4Qk3298Z
Lil Elon chibified by @Grok Imagine https://t.co/yw4Qk3298Z
Generation Models Know Space Unleashing Implicit 3D Priors for Scene Understanding paper: https://t.co/PH9c1fDAh6 https://t.co/60Xgl8Ffk1
Nvidia just released Nemotron-Cascade 2 on Hugging Face paper: https://t.co/ofx6zOlNic model: https://t.co/pgJht4jNE0
i'm sorry https://t.co/srrQA2h9VA
i'm sorry https://t.co/srrQA2h9VA
Monitoring the tituation... https://t.co/GbiLIgX1b2
Since its launch, there have been 60 million Copilot code reviews (and counting). π As AI speeds up how fast code ships, teams are using Copilot to keep review quality high without slowing down. Here's how we've implemented your feedback and evolved Copilot code reviews over time. β¬οΈ https://t.co/IDAi5TZaAQ
This is a cool, practical technique for increasing AI idea diversity by adding random priming phrases & bits of end words Similar prompts produce similar ideas, but since LLMs attend more to the start & end of inputs, this approach pushes towards novelty https://t.co/wwIPSRoFjs https://t.co/AcwGiCNzxo
Baidu releases SAMA on Hugging Face Factorizing instruction-guided video editing into semantic anchoring and motion alignment to achieve state-of-the-art performance among open-source models, remaining competitive with leading commercial systems like Kling-Omni. https://t.co/YHwYkWlHUK
