Resultados
38 artículosOrdenar
39:02ENAI’s product advantage is consistency. Its real weakness is getting conned.Presenta: Nathan LabenzNathan Labenz updates the Cognitive Revolution scouting report: jailbreaks, resume automation, Med-PaLM’s 40% radiology win, and what might follow Transformers.
1:32:22ENA Million Bytes and 256 Symbols: Meta's MEGABYTE Wants to Throw the Tokenizer AwayPresenta: Nathan LabenzLili Yu of Meta AI explains a multiscale transformer that predicts million-byte sequences, skips tokenization, and aims to model any file as raw bytes.
1:18:18ENFifteen Thousand Brain Numbers Were Enough to Reconstruct What Someone SawPresenta: Nathan LabenzTanishq Mathew Abraham walks through MindEye, the fMRI-to-image system that maps a handful of voxel values onto CLIP and diffusion models.
1:25:35ENTwo and a Half Years From 33.3% to 86.5%: Google Sent Foundation Models to Medical SchoolPresenta: Nathan LabenzVivek Natarajan walks Nathan Labenz through Med-PaLM and Med-PaLM 2: the data-light techniques, the nine-axis clinician evals, and why a high USMLE-style score is the wrong trophy.
1:29:31EN130,000 Teleop Demos: Google's GPT Playbook for RobotsPresenta: Nathan LabenzKeerthana Gopalakrishnan of Google Robotics tells Nathan Labenz how RT-1, SayCan, and PaLM-E turn language-model scaling into kitchen-ready mobile manipulators.
1:21:20ENFreeze the eyes and the brain. Train only the optic nerve.Presenta: Nathan LabenzJunnan Li and Dongxu Li tell Nathan Labenz how BLIP cleaned noisy web captions, and how BLIP-2 connected a frozen vision model to a frozen language model in under ten days on one A100.
1:23:18ENFrom GPT-2 to GPT-3 Moments in Robotics: Six DeepMind Papers on Bridging Robots and the InternetPresenta: Nathan LabenzKeerthana Gopalakrishnan and Ted Xiao walk through RT-2, RT-X, RT-Trajectory, AutoRT, Learning to Learn Faster and PIVOT, and say what is still missing for general-purpose robots.
1:53:14ENWorth Thousands, Priced at $20: Zvi Mowshowitz on OpenAI's Flurry, Gemini Bets, Deepfakes and a Compute CapPresenta: Nathan LabenzZvi Mowshowitz and Nathan Labenz survey OpenAI's launch spree, AI pricing and bundling, data pollution, deepfakes, the reversal curse, calls to cap training compute, and Amazon's Anthropic deal.
1:45:55ENFrom Static Structures to Moving Molecules: Where AI for Protein Design Stands After AlphaFold 3Presenta: Nathan LabenzComputational biochemist Amelie Schreiber walks Nathan Labenz through six months of progress in structure prediction, molecular dynamics, peptide and antibody models, and early agent-driven design workflows.
1:02:57ENVirtual Biopsies in One Second: How a Seven-Year-Old CycleGAN Is Beating Newer Models in the ClinicPresenta: Nathan LabenzTanishq Mathew Abraham explains how deep learning turns label-free 3D tissue images into virtual H&E stains fast enough to guide tumor surgery.
2:02:13ENWiring Without a Vision: What OpenAI's and Google's Dueling Launches Really SignalPresenta: Nathan LabenzZvi Mowshowitz and Nathan Labenz unpack the week of GPT-4o and Google I/O, from rushed demos and AI companions to the big tech singularity and the exodus from OpenAI's safety team.
1:48:13ENRobotics Hits Its GPT-3.5 Moment: Inside Gemini Robotics, From Embodied Reasoning to OrigamiPresenta: Nathan LabenzGoogle DeepMind researchers Keerthana Gopalakrishnan and Ted Xiao explain how Gemini Robotics works, what it can and cannot do, and where the field goes next.
1:17:31ENImage AI Is Still at the GPT-2 Stage: Suhail Doshi on Building a Unified Vision ModelPresenta: Nathan LabenzPlayground AI's founder explains why image generation has barely begun to deliver real utility, and why his team is betting 2024 on editing and a single model that can create, edit and understand pixels.
1:22:17ENOne Set of Weights, Fourteen Medical Tasks: Inside Google's Multimodal Med-PaLM MPresenta: Nathan LabenzVivek Natarajan and Tao Tu explain how Google built a single generalist biomedical model that reads images, genomics and text, and what it will take to put it in front of patients.









