Video wird geladen...

Video konnte nicht geladen werden

Zur Startseite

Introducing Marigold 🌼 - a universal monocular depth estimator, delivering incredibly sharp predictions in the wild! Based on Stable Diffusion, it is trained with synthetic depth data only and excels in zero-shot adaptation to real-world imagery. Check it out: 🌐 Website: 🤗 Hugging Face Space: 📄 Paper: 👾 Code:...

AntonObukhov1's profile picture

Anton Obukhov

2,702 subscribers

490,940 Aufrufe • vor 2 Jahren •via X (Twitter)

BildungWissenschaft & TechnologieKunst
LIVE

Anya Rossi• Live Now

Private livecam show

9 Kommentare

Profilbild von AI葵(Aoi) 🇹🇼NeRF技術型Vtuber
AI葵(Aoi) 🇹🇼NeRF技術型Vtubervor 2 Jahren

70s of inference for an image is very slow..

Profilbild von Anton Obukhov
Anton Obukhovvor 2 Jahren

We focus on the quality first - speed will come later through research and improvements in diffusion samplers. Arguably, Stable Diffusion is also slower than most of the GANs :)

Profilbild von Bilawal Sidhu
Bilawal Sidhuvor 2 Jahren

Damn those are some crispy depth maps!

Profilbild von Chris Offner
Chris Offnervor 2 Jahren

Wow, those depth maps look amazing. Great work! There goes my course project for Deep Learning. 😅

Profilbild von Boyuan Chen
Boyuan Chenvor 2 Jahren

Is this metric or relative depth?

Profilbild von Anton Obukhov
Anton Obukhovvor 2 Jahren

Relative

Profilbild von Janek Mann
Janek Mannvor 2 Jahren

I also consider this strong evidence that Stable Diffusion learned much stronger 3D scene understanding than commonly assumed.

Profilbild von Omar Sanseviero
Omar Sansevierovor 2 Jahren

Very very cool!

Profilbild von Tom
Tomvor 2 Jahren

Wow. More detailed than the ground truth, very cool to see.

Ähnliche Videos

🚀 VoxCPM 2 is live! 🎉 Another open-source AI #TTS model from China — and one that stands shoulder to shoulder with Qwen3-TTS, while bringing everything into a single unified model. After rapid iterations from V1 (zero-shot cloning) to V1.5 (long-form + fine-tuning), #VoxCPM has consistently pushed quality and usability forward. Now, VoxCPM 2 takes it further: 🔹30+ languages — truly global, truly local. 🔹Infinite voice design — type it, hear it, control it. From a whisper to a booming cinematic voice. 🔹Studio-grade audio — 48kHz ultra-high fidelity with emotional depth 🔹Diffusion-Autoregressive cloning — preserves more acoustic and emotional detail than token-based models like Qwen3-TTS 💡 Big shoutout to Grok — used your multi-image video magic for our launch demo. It’s scarily good at keeping visuals consistent across shots. Elon Elon Musk, this one’s for you. 😉 Check the demo & start cloning your dream voice: 🌐 Hugging Face Space: 🤗 Hugging Face Model: 🤖 ModelScope Model: 💻 GitHub： #TTS #AI #VoiceCloning #GrokImagine #ElonMusk #OpenBMB #VoxCPM
1:26

🚀 VoxCPM 2 is live! 🎉 Another open-source AI #TTS model from China — and one that stands shoulder to shoulder with Qwen3-TTS, while bringing everything into a single unified model. After rapid iterations from V1 (zero-shot cloning) to V1.5 (long-form + fine-tuning), #VoxCPM has consistently pushed quality and usability forward. Now, VoxCPM 2 takes it further: 🔹30+ languages — truly global, truly local. 🔹Infinite voice design — type it, hear it, control it. From a whisper to a booming cinematic voice. 🔹Studio-grade audio — 48kHz ultra-high fidelity with emotional depth 🔹Diffusion-Autoregressive cloning — preserves more acoustic and emotional detail than token-based models like Qwen3-TTS 💡 Big shoutout to Grok — used your multi-image video magic for our launch demo. It’s scarily good at keeping visuals consistent across shots. Elon Elon Musk, this one’s for you. 😉 Check the demo & start cloning your dream voice: 🌐 Hugging Face Space: 🤗 Hugging Face Model: 🤖 ModelScope Model: 💻 GitHub： #TTS #AI #VoiceCloning #GrokImagine #ElonMusk #OpenBMB #VoxCPM

OpenBMB

560,058 Aufrufe • vor 5 Monaten

I’m thrilled to announce that we just released GraspGen, a multi-year project we have been cooking at NVIDIA Robotics 🚀 GraspGen: A Diffusion-Based Framework for 6-DOF Grasping Grasping is a foundational challenge in robotics 🤖 — whether for industrial picking or general-purpose humanoids. VLA + real data collection is all the rage now but is expensive and scales poorly for this task. For every new gripper and/or scene, you’ll have to recollect the dataset in this paradigm for the best perf. 💡Key Idea: Since grasping is such a well-defined task in simulation - why can’t we just scale synthetic data generation and train a generative model for grasping? By embracing modularity and standardized grasp formats, we can make this a turnkey technology that works zero-shot for multiple settings. GraspGen is a modular framework for diffusion-based 6-DOF grasp generation that scales across embodiment types, observability conditions, clutter, task complexity. Key Features: ✅ Multi-embodiment support: suction, parallel-jaw, and multi-fingered grippers ✅ Generalization to partial + complete 3D point clouds ✅ Generalization to single-objects + cluttered scenes ✅ Modular design uses other robotics modules and foundation models (SAM2, cuRobo, FoundationStereo, FoundationPose). This allows GraspGen to focus on only one thing - grasp generation ✅ Training recipe: grasp discriminator is trained with On-Generator data from the diffusion model - so that it learns to correct the mistakes (if any) of the diffusion generator ✅ Real-time performance (~20 Hz) before any GPU acceleration; low memory footprint 📊 Results: • SOTA on the FetchBench [Han et al. CoRL 2024] benchmark • Zero-shot sim-to-real transfer on unknown objects and cluttered scenes • Dataset of 53M simulated grasps across 8K objects from Objaverse 📄 arXiv: 🌐 Website: 💻 Code: A huge thank you to everyone involved in this journey — excited to see what the community builds on top of it! Joint work with Clemens Eppner , Balakumar Sundaralingam , Yu-Wei, Jun Yamada Wentao Yuan and other collaborators #robotics #diffusionmodels #physicalAI #simtoreal
0:39

I’m thrilled to announce that we just released GraspGen, a multi-year project we have been cooking at NVIDIA Robotics 🚀 GraspGen: A Diffusion-Based Framework for 6-DOF Grasping Grasping is a foundational challenge in robotics 🤖 — whether for industrial picking or general-purpose humanoids. VLA + real data collection is all the rage now but is expensive and scales poorly for this task. For every new gripper and/or scene, you’ll have to recollect the dataset in this paradigm for the best perf. 💡Key Idea: Since grasping is such a well-defined task in simulation - why can’t we just scale synthetic data generation and train a generative model for grasping? By embracing modularity and standardized grasp formats, we can make this a turnkey technology that works zero-shot for multiple settings. GraspGen is a modular framework for diffusion-based 6-DOF grasp generation that scales across embodiment types, observability conditions, clutter, task complexity. Key Features: ✅ Multi-embodiment support: suction, parallel-jaw, and multi-fingered grippers ✅ Generalization to partial + complete 3D point clouds ✅ Generalization to single-objects + cluttered scenes ✅ Modular design uses other robotics modules and foundation models (SAM2, cuRobo, FoundationStereo, FoundationPose). This allows GraspGen to focus on only one thing - grasp generation ✅ Training recipe: grasp discriminator is trained with On-Generator data from the diffusion model - so that it learns to correct the mistakes (if any) of the diffusion generator ✅ Real-time performance (~20 Hz) before any GPU acceleration; low memory footprint 📊 Results: • SOTA on the FetchBench [Han et al. CoRL 2024] benchmark • Zero-shot sim-to-real transfer on unknown objects and cluttered scenes • Dataset of 53M simulated grasps across 8K objects from Objaverse 📄 arXiv: 🌐 Website: 💻 Code: A huge thank you to everyone involved in this journey — excited to see what the community builds on top of it! Joint work with Clemens Eppner , Balakumar Sundaralingam , Yu-Wei, Jun Yamada Wentao Yuan and other collaborators #robotics #diffusionmodels #physicalAI #simtoreal

Adithya Murali

24,296 Aufrufe • vor 1 Jahr

Today, we give robots a /skills library that self-evolves and compounds indefinitely! Introducing ASPIRE: a robot solving its 100th task is no longer as clueless as solving its first. Coding agents observe multimodal sensory traces from simulation and real robots, launch an evolutionary search over control programs, and distill the best know-how into an ever-expanding library. ASPIRE is a new type of continual learning: "training" is skill refinement instead of gradient descent. "Trained model" is a repo of sensorimotor skills instead of floating weights. “Distributed training” is a panel of agents each practicing a different skill instead of sharded minibatches. Here's the beauty: ASPIRE gives the tired terms "sim2real transfer" and "cross-embodiment transfer" a whole new meaning. Bridging the sim-to-real gap is notoriously brutal. An end-to-end policy has to swallow both the visual shift (sim looks toyish next to a real camera) and the subtle contact physics it never quite gets right. ASPIRE sidesteps the mess, because it doesn't ship pixels or weights across the gap, but ships the know-how. The robot still has to practice in the real world, not zero-shot, but it gets there way faster because it isn't rediscovering the strategy from scratch. Same for going single-arm to bimanual hardware, which usually requires new data and retraining from zero. ASPIRE achieves up to ~10x cut in "transfer learning” tokens (yes, tokens are the new unit of *training* compute ;) Check out our gallery of 150+ tasks and 90+ skills the robots taught themselves, all on the website! Kind of wild that we can ship the "learned weights" as an HTML page rather than a GGUF. We'll open-source the full stack so your own robot library starts compounding from ours! Deep dive in thread:
0:40

Today, we give robots a /skills library that self-evolves and compounds indefinitely! Introducing ASPIRE: a robot solving its 100th task is no longer as clueless as solving its first. Coding agents observe multimodal sensory traces from simulation and real robots, launch an evolutionary search over control programs, and distill the best know-how into an ever-expanding library. ASPIRE is a new type of continual learning: "training" is skill refinement instead of gradient descent. "Trained model" is a repo of sensorimotor skills instead of floating weights. “Distributed training” is a panel of agents each practicing a different skill instead of sharded minibatches. Here's the beauty: ASPIRE gives the tired terms "sim2real transfer" and "cross-embodiment transfer" a whole new meaning. Bridging the sim-to-real gap is notoriously brutal. An end-to-end policy has to swallow both the visual shift (sim looks toyish next to a real camera) and the subtle contact physics it never quite gets right. ASPIRE sidesteps the mess, because it doesn't ship pixels or weights across the gap, but ships the know-how. The robot still has to practice in the real world, not zero-shot, but it gets there way faster because it isn't rediscovering the strategy from scratch. Same for going single-arm to bimanual hardware, which usually requires new data and retraining from zero. ASPIRE achieves up to ~10x cut in "transfer learning” tokens (yes, tokens are the new unit of *training* compute ;) Check out our gallery of 150+ tasks and 90+ skills the robots taught themselves, all on the website! Kind of wild that we can ship the "learned weights" as an HTML page rather than a GGUF. We'll open-source the full stack so your own robot library starts compounding from ours! Deep dive in thread:

Jim Fan

211,649 Aufrufe • vor 2 Monaten

Check out our #ECCV2026 paper "Low-latency Event-based Object Detection with Spatially-Sparse Linear Attention", where we make linear attention sparse in space, recurrent in time, and parallel in training, enabling the first purely-linear-attention-based neural network for asynchronous object detection with #EventCameras, outperforming the previous best asynchronous method with 20x less computation with truly event-by-event inference on CPU! Code released! Paper: Code: Video: Event cameras promise extremely low-latency vision, but to fully exploit them, the neural network must be low-latency too. We introduce #SpatiallySparseLinearAttention (#SSLA) for asynchronous object detection directly from raw events. Linear attention is particularly appealing for event cameras: it can be trained efficiently in parallel on long event sequences, while at inference it operates recurrently, updating its prediction every time a new event arrives. The problem is that conventional linear attention updates its entire state for every event. For object detection, where fine spatial resolution matters, this quickly becomes expensive. Our key idea is simple: an event only carries information about a small spatial region, so why update the entire spatial state? SSLA updates only the relevant parts of the state, enabling fine-grained spatial representations while keeping per-event computation low. We achieve: - >20× lower per-event computation than the strongest prior asynchronous baseline - State-of-the-art accuracy among asynchronous object detection methods - Truly event-by-event inference on CPU, designed to preserve the latency advantage of event cameras Come to our poster on Friday September 11, 2026 from 4-6pm at ExHall #389 Reference: Haiqing Hao, Zhipeng Sui, Rong Zou, Zijia Dai, Nikola Zubić, Davide Scaramuzza, Wenhui Wang Low-latency Event-based Object Detection with Spatially-Sparse Linear Attention ECCV, 2026 Prophesee SynSense University of Zurich UZH Science European Research Council (ERC) UZHai UZH IfI Tesla BYD #EventCameras #ComputerVision #Robotics #DeepLearning #NeuromorphicVision #AI
2:40

Check out our #ECCV2026 paper "Low-latency Event-based Object Detection with Spatially-Sparse Linear Attention", where we make linear attention sparse in space, recurrent in time, and parallel in training, enabling the first purely-linear-attention-based neural network for asynchronous object detection with #EventCameras, outperforming the previous best asynchronous method with 20x less computation with truly event-by-event inference on CPU! Code released! Paper: Code: Video: Event cameras promise extremely low-latency vision, but to fully exploit them, the neural network must be low-latency too. We introduce #SpatiallySparseLinearAttention (#SSLA) for asynchronous object detection directly from raw events. Linear attention is particularly appealing for event cameras: it can be trained efficiently in parallel on long event sequences, while at inference it operates recurrently, updating its prediction every time a new event arrives. The problem is that conventional linear attention updates its entire state for every event. For object detection, where fine spatial resolution matters, this quickly becomes expensive. Our key idea is simple: an event only carries information about a small spatial region, so why update the entire spatial state? SSLA updates only the relevant parts of the state, enabling fine-grained spatial representations while keeping per-event computation low. We achieve: - >20× lower per-event computation than the strongest prior asynchronous baseline - State-of-the-art accuracy among asynchronous object detection methods - Truly event-by-event inference on CPU, designed to preserve the latency advantage of event cameras Come to our poster on Friday September 11, 2026 from 4-6pm at ExHall #389 Reference: Haiqing Hao, Zhipeng Sui, Rong Zou, Zijia Dai, Nikola Zubić, Davide Scaramuzza, Wenhui Wang Low-latency Event-based Object Detection with Spatially-Sparse Linear Attention ECCV, 2026 Prophesee SynSense University of Zurich UZH Science European Research Council (ERC) UZHai UZH IfI Tesla BYD #EventCameras #ComputerVision #Robotics #DeepLearning #NeuromorphicVision #AI

Davide Scaramuzza

51,571 Aufrufe • vor 13 Tagen

Introducing LifeGPT, showing that LLMs can simulate complex, Turing-complete systems like Conway's Game of Life with near-perfect accuracy—no prior topology needed.🌐This unlocks new potential for AI in modeling self-organizing systems in biology, materials science, & beyond.🔬🤖 #AI #LifeGPT. Cellular Automata (CA), like Conway's Game of Life ("Life"), are computationally irreducible, meaning their evolution is difficult to predict without an a-priori understanding of the rules of the game, including the topology on which it is played. LifeGPT is a topology-agnostic generative model that learns the rules of Life without prior knowledge of its grid structure or boundary conditions, from only a tiny number of game states. The success in simulating Life suggests promising avenues for scientific discovery, particularly in bridging the gap between AI, artificial life, and real-world biological systems, for both forward and inverse problems. The potential for universal computation within generative AI, including LLMs, through approaches like LifeGPT, represents an exciting area for future research, especially when combined with reinforcement learning. Model Convergence: LifeGPT exhibits rapid convergence during training, achieving high accuracy in predicting next-game-states. We attribute the non-zero cross-entropy loss to the lack of causal relationships within randomly generated ICs. Accuracy & Temperature: LifeGPT achieves near-perfect accuracy, particularly at lower sampling temperatures, but can be continually tuned towards higher creativity to discover patterns that the original ruleset would not be able to produce. This finding highlights the trade-off between model creativity (higher temperature) and accuracy in deterministic predictions, with high relevance to model real-world dynamical systems for which no closed-form rulesets exist. Zero/Few-Shot Learning: Trained on a small fraction of possible initial conditions, LifeGPT demonstrates strong zero/few-shot learning, accurately simulating Life for unseen initial conditions. However, rare prediction errors highlight that LifeGPT approximates rather than perfectly replicates the Life algorithm. Autoregressive Autoregressor: A recursive implementation of LifeGPT demonstrates the model's ability to simulate Life over multiple timesteps. LifeGPT is topology-agnostic with respect to its training data and our results show that a GPT model is capable of capturing the deterministic rules of a Turing-complete system with near-perfect accuracy, given sufficiently diverse training data. The work showcases the possibility for future models to synthesize stochastic generative capabilities with deterministic computational capabilities. Link to code, paper, etc. below. Podcast generated using #NotebookLM. LAMM@MIT DMSE at MIT
9:59

Introducing LifeGPT, showing that LLMs can simulate complex, Turing-complete systems like Conway's Game of Life with near-perfect accuracy—no prior topology needed.🌐This unlocks new potential for AI in modeling self-organizing systems in biology, materials science, & beyond.🔬🤖 #AI #LifeGPT. Cellular Automata (CA), like Conway's Game of Life ("Life"), are computationally irreducible, meaning their evolution is difficult to predict without an a-priori understanding of the rules of the game, including the topology on which it is played. LifeGPT is a topology-agnostic generative model that learns the rules of Life without prior knowledge of its grid structure or boundary conditions, from only a tiny number of game states. The success in simulating Life suggests promising avenues for scientific discovery, particularly in bridging the gap between AI, artificial life, and real-world biological systems, for both forward and inverse problems. The potential for universal computation within generative AI, including LLMs, through approaches like LifeGPT, represents an exciting area for future research, especially when combined with reinforcement learning. Model Convergence: LifeGPT exhibits rapid convergence during training, achieving high accuracy in predicting next-game-states. We attribute the non-zero cross-entropy loss to the lack of causal relationships within randomly generated ICs. Accuracy & Temperature: LifeGPT achieves near-perfect accuracy, particularly at lower sampling temperatures, but can be continually tuned towards higher creativity to discover patterns that the original ruleset would not be able to produce. This finding highlights the trade-off between model creativity (higher temperature) and accuracy in deterministic predictions, with high relevance to model real-world dynamical systems for which no closed-form rulesets exist. Zero/Few-Shot Learning: Trained on a small fraction of possible initial conditions, LifeGPT demonstrates strong zero/few-shot learning, accurately simulating Life for unseen initial conditions. However, rare prediction errors highlight that LifeGPT approximates rather than perfectly replicates the Life algorithm. Autoregressive Autoregressor: A recursive implementation of LifeGPT demonstrates the model's ability to simulate Life over multiple timesteps. LifeGPT is topology-agnostic with respect to its training data and our results show that a GPT model is capable of capturing the deterministic rules of a Turing-complete system with near-perfect accuracy, given sufficiently diverse training data. The work showcases the possibility for future models to synthesize stochastic generative capabilities with deterministic computational capabilities. Link to code, paper, etc. below. Podcast generated using #NotebookLM. LAMM@MIT DMSE at MIT

Markus J. Buehler

114,251 Aufrufe • vor 1 Jahr

Excited to share TERRA, a tissue world model 🧬 Over ~1.5 years we ran a large data-generation + modelling effort to build a world model for human tissues, pretrained on 112M cells from spatial transcriptomics (mostly Xenium 5000-plex + public data). It's built on one of the largest human spatial transcriptomics corpora assembled to date, spanning 20 tissues across development, health and 26 disease conditions, ~two-thirds newly generated in-house. Why a "world model" for tissue? Images have universal representations (ViT/DINOv3), so do proteins (ESM, Alex Rives) and pathology (UNI, Faisal Mahmood). We've worked hard to build something similar for human tissue: one model that captures its multi-scale logic, genes → cells → their native microenvironments. Like the JEPA approach Yann LeCun has championed, TERRA learns by prediction in embedding space, but for human tissue. How it works: it tokenises each cell together with its nearest neighbours into one sequence while keeping every gene's identity, then masks part of a neighbourhood and predicts the representation of the hidden part, not raw noisy counts. From one backbone it reads out three scales, gene embeddings (what a gene is doing in a cell and its niche), cell embeddings (cell type and state) and neighbourhood embeddings (the niche), and because it keeps gene-level resolution it can knock a gene out in silico and predict the response. Applied entirely zero-shot, TERRA maps and perturbs human tissue across unseen organs, diseases and technologies, outperforming existing spatial approaches. Three take-homes: 1️⃣ One model, any tissue. A single pretrained backbone provides tissue representations zero-shot, handling genes, cells and niches across organs and platforms, off the shelf. 2️⃣ New biology, development to clinic. We built a new spatial atlas of the developing human pancreas and found an islet-associated capillary state that looks like a precursor of mature islet vasculature. In kidney, TERRA's in silico knockouts predicted the tissue-injury programme from cancer immunotherapy (checkpoint blockade), confirmed in treated kidneys, detected in blood, and linked to declining kidney function. 3️⃣ A grammar of tissue architecture. By coupling each cell's state to its niche, TERRA defines recurring cross-organ "archetypes" of macrophage neighbourhoods, including a tumour-boundary niche that tracks poor survival in kidney cancer. TERRA is already in use: it powered our recent skin atlas of hidden immune-memory niches ( with more studies coming soon. This was an amazing collaboration between clinicians, machine-learning scientists and cell biologists 🙏 Led by Sebastian Birk, Vali Sanian Amirhosein Vahidi, Samuel Ogden, Daniyal Jafree, Adib Miraki, Carlo Leonardi and Arpit Merchant, with Lassi Paavolainen, Menna Clatworthy, Omer Ali Bayraktar, Muzlifah Haniffa, Tom Mitchell and Mostafa Bakhti. Huge thanks too to everyone who shared data and helped along the way. What excites me most is seeing how the community builds on this. The model, code and tutorials are all public, so anyone can run TERRA on their own tissues, extend it, or build new models on top. Huge thanks to the whole team across Wellcome Sanger Institute and our many collaborators. 📄 Paper: 💻 Code: 🤗 Model: #SpatialTranscriptomics #SpatialGenomics #FoundationModels #AI4Science #MachineLearning #ComputationalBiology #SingleCell #WorldModels
2:15

Excited to share TERRA, a tissue world model 🧬 Over ~1.5 years we ran a large data-generation + modelling effort to build a world model for human tissues, pretrained on 112M cells from spatial transcriptomics (mostly Xenium 5000-plex + public data). It's built on one of the largest human spatial transcriptomics corpora assembled to date, spanning 20 tissues across development, health and 26 disease conditions, ~two-thirds newly generated in-house. Why a "world model" for tissue? Images have universal representations (ViT/DINOv3), so do proteins (ESM, Alex Rives) and pathology (UNI, Faisal Mahmood). We've worked hard to build something similar for human tissue: one model that captures its multi-scale logic, genes → cells → their native microenvironments. Like the JEPA approach Yann LeCun has championed, TERRA learns by prediction in embedding space, but for human tissue. How it works: it tokenises each cell together with its nearest neighbours into one sequence while keeping every gene's identity, then masks part of a neighbourhood and predicts the representation of the hidden part, not raw noisy counts. From one backbone it reads out three scales, gene embeddings (what a gene is doing in a cell and its niche), cell embeddings (cell type and state) and neighbourhood embeddings (the niche), and because it keeps gene-level resolution it can knock a gene out in silico and predict the response. Applied entirely zero-shot, TERRA maps and perturbs human tissue across unseen organs, diseases and technologies, outperforming existing spatial approaches. Three take-homes: 1️⃣ One model, any tissue. A single pretrained backbone provides tissue representations zero-shot, handling genes, cells and niches across organs and platforms, off the shelf. 2️⃣ New biology, development to clinic. We built a new spatial atlas of the developing human pancreas and found an islet-associated capillary state that looks like a precursor of mature islet vasculature. In kidney, TERRA's in silico knockouts predicted the tissue-injury programme from cancer immunotherapy (checkpoint blockade), confirmed in treated kidneys, detected in blood, and linked to declining kidney function. 3️⃣ A grammar of tissue architecture. By coupling each cell's state to its niche, TERRA defines recurring cross-organ "archetypes" of macrophage neighbourhoods, including a tumour-boundary niche that tracks poor survival in kidney cancer. TERRA is already in use: it powered our recent skin atlas of hidden immune-memory niches ( with more studies coming soon. This was an amazing collaboration between clinicians, machine-learning scientists and cell biologists 🙏 Led by Sebastian Birk, Vali Sanian Amirhosein Vahidi, Samuel Ogden, Daniyal Jafree, Adib Miraki, Carlo Leonardi and Arpit Merchant, with Lassi Paavolainen, Menna Clatworthy, Omer Ali Bayraktar, Muzlifah Haniffa, Tom Mitchell and Mostafa Bakhti. Huge thanks too to everyone who shared data and helped along the way. What excites me most is seeing how the community builds on this. The model, code and tutorials are all public, so anyone can run TERRA on their own tissues, extend it, or build new models on top. Huge thanks to the whole team across Wellcome Sanger Institute and our many collaborators. 📄 Paper: 💻 Code: 🤗 Model: #SpatialTranscriptomics #SpatialGenomics #FoundationModels #AI4Science #MachineLearning #ComputationalBiology #SingleCell #WorldModels

Mo Lotfollahi

50,890 Aufrufe • vor 1 Monat

Here is a live demo of our AI solution I've been building non-stop over the past 8 months Binary Defense. How it works: Our own model trained on our analysts behavior. Our analysts submit tickets as false positives/true positives with context which enriches our LLM to be smarter over time. Key Highlights: If its a binary - will automatically spin up an agent for reverse engineering it and using EMBER ML to understand behavior and intent of the binary. File formats: Supports a vast array of pretty much any filetype, including email attachments like SVG, LNK, etc. Can handle DLLs, ELF, EXEs, PDF, XLS, DOC, etc. Interrogates the full chain of all events irrespective of log sources. Can handle any format of logs and integrates into APIs of customers for additional agentic data looping for confidence ranking when needed. This is an example of the back-end UI, this is transparent to analysts and enriches the alarms automatically in our SOAR. In these examples there's three different types: 1. Regsvr32 + sct downloader + scrobj.dll code execution - checks reputation of domain, pulls in threat intel, looks at entire picture of the chain - downloads the file itself and inspects for code analysis. Determines if malicious as well as historically looking back if seen in customer before in past. 2. Powershell Obfuscation - uses a universal decoder to un-obfuscate powershell and look at the raw code. Can handle pretty much any obfuscation thrown at it (thanks Justin Elze). 3. Email with malicious SVG - checks tonality of email, are they creating urgency to take action (increases confidence) - disassembles SVG to understand malicious content - checks URL to determine if harvesting credentials, payload delivery, etc. Creates an entire kill chain analysis with full response and dissecting of the attack to the analyst in seconds. Has greatly sped up our ability to respond to incidents and allowing analysts to focus on the most important alarms through prioritization. Once cool thing I've worked heavily on is a synthetic data normalizer which when an analyst says "Yes this is bad with context" or "No this is a false positive" - our local model generates training data to be smarter in the future without using the actual customer data to train it. The customers actual data is immediately destroyed once training data off of the original alarm is generated and contains no customer-centric data at all. We also have three model tiers. Opt-In (collective model, again no customer data but every organization contributes to training). Opt-Out - does not train on any customer data for customers who opt-out. Private LLM - LLM created specifically for individual customer and trains only off of their data. Uses shared model collective for better confidence rankings. It will generate automated playbooks to run based on confidence rankings to take action on behalf of the customer. Still human driven on execution - has to approve playbook actions. This thing is cooking and so cool to see this work live and shut down attackers much faster! If confidence ranking is low - will automatically attempt to enrich data through customer environments for better confidence rankings. Additionally if the model isn't trained well on a certain technology, I have created something we call "Nexus" that will research new protocols, devices, SDKs, etc and generate training data automatically. Works well for zero-days for example, point to a tweet, or a research paper, and automatically generates training data to recognize this attack much faster. Have over 8000+ yara rule integrations that help with confidence boosting as well that is automatically incorporated into the analysis. Creating some amazing stuff at Binary Defense that isn't marketing fluff - actionable things that are making a huge difference in this industry. #BinaryDefense
0:32

Here is a live demo of our AI solution I've been building non-stop over the past 8 months Binary Defense. How it works: Our own model trained on our analysts behavior. Our analysts submit tickets as false positives/true positives with context which enriches our LLM to be smarter over time. Key Highlights: If its a binary - will automatically spin up an agent for reverse engineering it and using EMBER ML to understand behavior and intent of the binary. File formats: Supports a vast array of pretty much any filetype, including email attachments like SVG, LNK, etc. Can handle DLLs, ELF, EXEs, PDF, XLS, DOC, etc. Interrogates the full chain of all events irrespective of log sources. Can handle any format of logs and integrates into APIs of customers for additional agentic data looping for confidence ranking when needed. This is an example of the back-end UI, this is transparent to analysts and enriches the alarms automatically in our SOAR. In these examples there's three different types: 1. Regsvr32 + sct downloader + scrobj.dll code execution - checks reputation of domain, pulls in threat intel, looks at entire picture of the chain - downloads the file itself and inspects for code analysis. Determines if malicious as well as historically looking back if seen in customer before in past. 2. Powershell Obfuscation - uses a universal decoder to un-obfuscate powershell and look at the raw code. Can handle pretty much any obfuscation thrown at it (thanks Justin Elze). 3. Email with malicious SVG - checks tonality of email, are they creating urgency to take action (increases confidence) - disassembles SVG to understand malicious content - checks URL to determine if harvesting credentials, payload delivery, etc. Creates an entire kill chain analysis with full response and dissecting of the attack to the analyst in seconds. Has greatly sped up our ability to respond to incidents and allowing analysts to focus on the most important alarms through prioritization. Once cool thing I've worked heavily on is a synthetic data normalizer which when an analyst says "Yes this is bad with context" or "No this is a false positive" - our local model generates training data to be smarter in the future without using the actual customer data to train it. The customers actual data is immediately destroyed once training data off of the original alarm is generated and contains no customer-centric data at all. We also have three model tiers. Opt-In (collective model, again no customer data but every organization contributes to training). Opt-Out - does not train on any customer data for customers who opt-out. Private LLM - LLM created specifically for individual customer and trains only off of their data. Uses shared model collective for better confidence rankings. It will generate automated playbooks to run based on confidence rankings to take action on behalf of the customer. Still human driven on execution - has to approve playbook actions. This thing is cooking and so cool to see this work live and shut down attackers much faster! If confidence ranking is low - will automatically attempt to enrich data through customer environments for better confidence rankings. Additionally if the model isn't trained well on a certain technology, I have created something we call "Nexus" that will research new protocols, devices, SDKs, etc and generate training data automatically. Works well for zero-days for example, point to a tweet, or a research paper, and automatically generates training data to recognize this attack much faster. Have over 8000+ yara rule integrations that help with confidence boosting as well that is automatically incorporated into the analysis. Creating some amazing stuff at Binary Defense that isn't marketing fluff - actionable things that are making a huge difference in this industry. #BinaryDefense

Dave Kennedy

29,036 Aufrufe • vor 6 Monaten

We’re open sourcing the first document OCR benchmark for the agentic era, ParseBench. Document parsing is the foundation of every AI agent that works with real-world files. ParseBench is a benchmark that measures parsing quality specifically for agent knowledge work: ✅ It optimizes for semantic correctness (instead of exact similarity) ✅ It has the most comprehensive distribution of real-world enterprise documents It contains ~2,000 human-verified enterprise document pages with 167,000+ test rules across five dimensions that matter most: tables, charts, content faithfulness, semantic formatting, and visual grounding. We benchmarked 14 known document parsers on ParseBench, from frontier/OSS VLMs to specialized parsers to LlamaParse. Here are some of our findings: 💡 Increasing compute budget yields diminishing returns - Gemini/gpt-5-mini/haiku gain 3-5 points from minimal to high thinking, at 4x the cost. 💡 Charts are the most polarizing dimension for evaluation. Most specialized parsers score below 6%, while some VLM-based parsers do a bit better. 💡 VLMs are great at visual understanding but terrible at layout extraction. GPT-5-mini/haiku score below 10% on our visual grounding task, all specialized parsers do much better. 💡 No method crushes all 5 dimensions at once, but LlamaParse achieves the highest overall score at 84.9%, and is the leader in 4 out of the 5 dimensions. This is by far the deepest technical work that we’ve published as a company. I would encourage you to start with our blog and explore our links to Hugging Face to GitHub. All the details are in our full 35-page (!!) ArXiv whitepaper. 🌐: Blog: 📄 Paper: 💻 Code: 📊 Dataset: 🎥 YouTube:
1:08

We’re open sourcing the first document OCR benchmark for the agentic era, ParseBench. Document parsing is the foundation of every AI agent that works with real-world files. ParseBench is a benchmark that measures parsing quality specifically for agent knowledge work: ✅ It optimizes for semantic correctness (instead of exact similarity) ✅ It has the most comprehensive distribution of real-world enterprise documents It contains ~2,000 human-verified enterprise document pages with 167,000+ test rules across five dimensions that matter most: tables, charts, content faithfulness, semantic formatting, and visual grounding. We benchmarked 14 known document parsers on ParseBench, from frontier/OSS VLMs to specialized parsers to LlamaParse. Here are some of our findings: 💡 Increasing compute budget yields diminishing returns - Gemini/gpt-5-mini/haiku gain 3-5 points from minimal to high thinking, at 4x the cost. 💡 Charts are the most polarizing dimension for evaluation. Most specialized parsers score below 6%, while some VLM-based parsers do a bit better. 💡 VLMs are great at visual understanding but terrible at layout extraction. GPT-5-mini/haiku score below 10% on our visual grounding task, all specialized parsers do much better. 💡 No method crushes all 5 dimensions at once, but LlamaParse achieves the highest overall score at 84.9%, and is the leader in 4 out of the 5 dimensions. This is by far the deepest technical work that we’ve published as a company. I would encourage you to start with our blog and explore our links to Hugging Face to GitHub. All the details are in our full 35-page (!!) ArXiv whitepaper. 🌐: Blog: 📄 Paper: 💻 Code: 📊 Dataset: 🎥 YouTube:

Jerry Liu

108,093 Aufrufe • vor 4 Monaten

Introducing Sharpe Search: On-Chain Search AI Agent Powered by Hive Intelligence We’re thrilled to announce the launch of Sharpe Search, a crypto search AI agent powered by Hive Intelligence Designed to simplify blockchain data interaction, Sharpe Search represents a significant step toward making crypto more accessible and actionable for users at every level. Sharpe Search leverages Hive Intelligence’s advanced search API to provide real-time, actionable insights across the blockchain ecosystem. Here’s a detailed look at what Sharpe Search is, how it works: What Is Sharpe Search? At its core, Sharpe Search is an AI agent purpose-built for querying and analyzing on-chain data. It takes the complexity out of blockchain exploration by enabling users to ask questions in plain language and receive detailed, accurate responses. Whether you’re looking to monitor wallet activity, track portfolio positions, or analyze transaction history, Sharpe Search ensures that the answers are at your fingertips—accurate, comprehensive, and delivered instantly. How Does Sharpe Search Work? Sharpe Search is powered by Hive Intelligence, a search engine API designed to make blockchain data easily accessible and AI-ready. Here’s a breakdown of how it enables Sharpe Search to function effectively: 1. LLM-Optimized Query Processing Sharpe Search leverages Hive Intelligence's optimized responses for large language models. This ensures that AI agents can process blockchain data in a structured format, delivering precise answers to complex user queries. 2. Natural Language Interaction Forget the need for technical knowledge. Sharpe Search supports natural language queries, making it as simple as typing: - “What tokens are in my wallet? Am I eligible for any airdrop I haven't claimed yet?” - “Check me my last 100 transactions, tell me if I interacted with any protocol with recent hacks” - “Track my wallet activity over the past month, suggest optimised portfolio based on best stable yields available” 3. Real-Time Insights Across Multi-Chains Using Hive Intelligence, Sharpe Search connects to over 20 chains and 5000+ Protocols. This real-time access ensures that the AI agent provides up-to-date and actionable insights, no matter how dynamic the blockchain environment. 4. Unified API Access Sharpe Search consolidates fragmented blockchain data through Hive’s unified API. Instead of dealing with multiple integrations, Sharpe Search uses a single access point to aggregate and query data, reducing complexity for both users and developers. Technical Depth: The AI Agent Advantage Sharpe Search's design philosophy revolves around the principle of creating an intuitive, AI-driven experience. Here’s what makes its technology stand out: Data Indexing and Aggregation: Hive Intelligence employs advanced indexing algorithms to aggregate data from multiple chains. This ensures that Sharpe Search can retrieve information within milliseconds, even when querying vast datasets. Dynamic Updates: Blockchain data is volatile. Sharpe Search processes dynamic updates in real time, enabling users to act on the most recent metrics, transactions, and balances without delays. Contextual Understanding: The AI agent parses natural language queries and contextualizes them to blockchain-specific scenarios. For instance, when querying “Show portfolio details,” Sharpe Search understands the underlying requirements—fetching wallet holdings, token values, and current positions. Hive Intelligence: The Backbone of Sharpe Search While Sharpe Search takes center stage, Hive Intelligence provides the critical infrastructure to make it all possible. Its LLM-ready responses and multi-chain support ensure that Sharpe Search operates at the forefront of blockchain data accessibility. By launching Hive Intelligence through Sharpe Launchpad, Sharpe reinforces its commitment to supporting innovation in the blockchain space. Hive’s infrastructure not only powers Sharpe Search but also lays the groundwork for future AI agents to thrive in the ecosystem. What’s Next for Sharpe Search? Currently in invite-only access, Sharpe Search is preparing for a broader public release. Future updates will include: - Expanded Blockchain Coverage: More chains and protocols will be added. - Enhanced Query Flexibility: Even more advanced natural language capabilities. Stay tuned for the public launch and get ready to explore crypto like never before!
1:33

Introducing Sharpe Search: On-Chain Search AI Agent Powered by Hive Intelligence We’re thrilled to announce the launch of Sharpe Search, a crypto search AI agent powered by Hive Intelligence Designed to simplify blockchain data interaction, Sharpe Search represents a significant step toward making crypto more accessible and actionable for users at every level. Sharpe Search leverages Hive Intelligence’s advanced search API to provide real-time, actionable insights across the blockchain ecosystem. Here’s a detailed look at what Sharpe Search is, how it works: What Is Sharpe Search? At its core, Sharpe Search is an AI agent purpose-built for querying and analyzing on-chain data. It takes the complexity out of blockchain exploration by enabling users to ask questions in plain language and receive detailed, accurate responses. Whether you’re looking to monitor wallet activity, track portfolio positions, or analyze transaction history, Sharpe Search ensures that the answers are at your fingertips—accurate, comprehensive, and delivered instantly. How Does Sharpe Search Work? Sharpe Search is powered by Hive Intelligence, a search engine API designed to make blockchain data easily accessible and AI-ready. Here’s a breakdown of how it enables Sharpe Search to function effectively: 1. LLM-Optimized Query Processing Sharpe Search leverages Hive Intelligence's optimized responses for large language models. This ensures that AI agents can process blockchain data in a structured format, delivering precise answers to complex user queries. 2. Natural Language Interaction Forget the need for technical knowledge. Sharpe Search supports natural language queries, making it as simple as typing: - “What tokens are in my wallet? Am I eligible for any airdrop I haven't claimed yet?” - “Check me my last 100 transactions, tell me if I interacted with any protocol with recent hacks” - “Track my wallet activity over the past month, suggest optimised portfolio based on best stable yields available” 3. Real-Time Insights Across Multi-Chains Using Hive Intelligence, Sharpe Search connects to over 20 chains and 5000+ Protocols. This real-time access ensures that the AI agent provides up-to-date and actionable insights, no matter how dynamic the blockchain environment. 4. Unified API Access Sharpe Search consolidates fragmented blockchain data through Hive’s unified API. Instead of dealing with multiple integrations, Sharpe Search uses a single access point to aggregate and query data, reducing complexity for both users and developers. Technical Depth: The AI Agent Advantage Sharpe Search's design philosophy revolves around the principle of creating an intuitive, AI-driven experience. Here’s what makes its technology stand out: Data Indexing and Aggregation: Hive Intelligence employs advanced indexing algorithms to aggregate data from multiple chains. This ensures that Sharpe Search can retrieve information within milliseconds, even when querying vast datasets. Dynamic Updates: Blockchain data is volatile. Sharpe Search processes dynamic updates in real time, enabling users to act on the most recent metrics, transactions, and balances without delays. Contextual Understanding: The AI agent parses natural language queries and contextualizes them to blockchain-specific scenarios. For instance, when querying “Show portfolio details,” Sharpe Search understands the underlying requirements—fetching wallet holdings, token values, and current positions. Hive Intelligence: The Backbone of Sharpe Search While Sharpe Search takes center stage, Hive Intelligence provides the critical infrastructure to make it all possible. Its LLM-ready responses and multi-chain support ensure that Sharpe Search operates at the forefront of blockchain data accessibility. By launching Hive Intelligence through Sharpe Launchpad, Sharpe reinforces its commitment to supporting innovation in the blockchain space. Hive’s infrastructure not only powers Sharpe Search but also lays the groundwork for future AI agents to thrive in the ecosystem. What’s Next for Sharpe Search? Currently in invite-only access, Sharpe Search is preparing for a broader public release. Future updates will include: - Expanded Blockchain Coverage: More chains and protocols will be added. - Enhanced Query Flexibility: Even more advanced natural language capabilities. Stay tuned for the public launch and get ready to explore crypto like never before!

Sharpe AI

263,308 Aufrufe • vor 1 Jahr

Introducing BioCLIP: A Vision Foundation Model for the Tree of Life A foundation model that strongly generalizes on the tree of life (2M+ species), outperforming OpenAI CLIP by 18% in zero-shot classification, and supports open-ended classification over almost the entire tree of life What's the secrete ingredients? > Data: we curate and release TreeOfLife-10M, the largest and most diverse ML-ready dataset of organism images to date. It contains 10.4M images for over 450K taxa, sourced from iNaturalist, BIOSCAN, and Encyclopedia of Life. > Modeling: we creatively repurposes CLIP's multimodal contrastive learning objective for hierarchical image classification. The autoregressive language model naturally encodes the hierarchy of the tree of life taxonomy, which in turn bakes the hierarchical representation into the vision transformer encoder. Key results > Strong zero/few-shot classification for animals/plants/fungi, including rare species, outperforming CLIP by avg 16-18% absolute. > T-sne visualization shows that BioCLIP's vision encoder has captued the fine-grained hierarchical structure of the tree of life > BioCLIP is a kind of universal classifier for the tree of life. Just give it an organism image and it will likely find the correct species (among top 5)! But use it with caution; it's not perfect yet.. Final remarks > AI for Science is really hard but extremely rewarding! It took us a ton of time (1+ year) and frustration trying to find a plausible way to integrate the tree of life taxonomy into foundation model training. But when the "Eureka!" moment came and the idea hit us (by the great Wei-Lun Chao) that CLIP's multimodal contrastive learning objective can be repurposed for that, everything just follows naturally. It was truly a moment of joy and excitement! > BioCLIP is our first attempt at foundation models for biology, but it certainly won't be the last! There's so much more to do at the intersection of one of the oldest scientific disciplines and the young but thriving field of AI. Biological intelligence is the foundation for artificial intelligence, and artificial intelligence will in turn become the most important tool for us to unraval the mysteries of biological intelligence. We are hiring postdocs and PhDs in the NSF Imageomics Institute institute to explore this exciting field! Drop us an email. also happy to chat about it at #NeurIPS2023 with any of Tanya, Wei-Lun Chao, or me. - paper: - project: - demo: - model: - data (TreeOfLife-10M): to be released on Hugging Face soon joint work with the amazing Imageomics Institute team: @samstevens6860 Lisa Wu, Matt Thompson, Elizabeth Campolongo Chan Hee (Luke) Song David Carlyn Li Dong Wasila Dahdul Chuck Stewart, Tanya Berger-Wolf Wei-Lun Chao Yu Su
0:31

Introducing BioCLIP: A Vision Foundation Model for the Tree of Life A foundation model that strongly generalizes on the tree of life (2M+ species), outperforming OpenAI CLIP by 18% in zero-shot classification, and supports open-ended classification over almost the entire tree of life What's the secrete ingredients? > Data: we curate and release TreeOfLife-10M, the largest and most diverse ML-ready dataset of organism images to date. It contains 10.4M images for over 450K taxa, sourced from iNaturalist, BIOSCAN, and Encyclopedia of Life. > Modeling: we creatively repurposes CLIP's multimodal contrastive learning objective for hierarchical image classification. The autoregressive language model naturally encodes the hierarchy of the tree of life taxonomy, which in turn bakes the hierarchical representation into the vision transformer encoder. Key results > Strong zero/few-shot classification for animals/plants/fungi, including rare species, outperforming CLIP by avg 16-18% absolute. > T-sne visualization shows that BioCLIP's vision encoder has captued the fine-grained hierarchical structure of the tree of life > BioCLIP is a kind of universal classifier for the tree of life. Just give it an organism image and it will likely find the correct species (among top 5)! But use it with caution; it's not perfect yet.. Final remarks > AI for Science is really hard but extremely rewarding! It took us a ton of time (1+ year) and frustration trying to find a plausible way to integrate the tree of life taxonomy into foundation model training. But when the "Eureka!" moment came and the idea hit us (by the great Wei-Lun Chao) that CLIP's multimodal contrastive learning objective can be repurposed for that, everything just follows naturally. It was truly a moment of joy and excitement! > BioCLIP is our first attempt at foundation models for biology, but it certainly won't be the last! There's so much more to do at the intersection of one of the oldest scientific disciplines and the young but thriving field of AI. Biological intelligence is the foundation for artificial intelligence, and artificial intelligence will in turn become the most important tool for us to unraval the mysteries of biological intelligence. We are hiring postdocs and PhDs in the NSF Imageomics Institute institute to explore this exciting field! Drop us an email. also happy to chat about it at #NeurIPS2023 with any of Tanya, Wei-Lun Chao, or me. - paper: - project: - demo: - model: - data (TreeOfLife-10M): to be released on Hugging Face soon joint work with the amazing Imageomics Institute team: @samstevens6860 Lisa Wu, Matt Thompson, Elizabeth Campolongo Chan Hee (Luke) Song David Carlyn Li Dong Wasila Dahdul Chuck Stewart, Tanya Berger-Wolf Wei-Lun Chao Yu Su

Yu Su

80,660 Aufrufe • vor 2 Jahren

🚨They CANNOT SCAN kilometers into SOLID ROCK! That is what skeptics of the Khafre Project’s findings say, and they are correct to point out that Synthetic Aperture Radar (SAR) can't do this. They are also TOTALLY MISTAKEN, here is why: The Khafre team are not using conventional SAR imaging in the way critics assume. They are using a patented form of SAR Doppler Tomography, pioneered by Prof. Filippo Biondi. This is not simple surface imaging. It is a phase-coherent interferometric method that detects subtle Doppler frequency shifts caused by internal micro-vibrations within dense structures. Instead of trying to penetrate rock, it “listens” to tiny seismic vibrations in the stone. Biondi’s trick is to capture micro-motions. Tiny seismic or structural tremors slightly shift the radar’s frequency (Doppler effect). By analyzing these Doppler shifts across multiple SAR images, they can reconstruct a 3D tomographic image of what’s inside, like a CT-scan from space. Prof. Filippop Biondi's patent (PCT/EP2023/064345) explicitly describes processing “coherent vibrational Doppler information” in SAR to allow penetrating 3D imaging "over a depth of several kilometers". In other words, it effectively turns the radar into a spaceborne sonar, using Earth’s natural vibrations to “sound” the subsurface, something ordinary SAR can’t do. A peer-reviewed Remote Sensing paper describes using COSMO-SkyMed SAR data to map new shafts and chambers inside Khufu . This case study in a scientific journal shows the technique in action (with high-res 3D results!). Beyond pyramids, the technique has practical uses. For bridges and infrastructure, Biondi’s SAR Doppler method can extract a structure’s “vibration profile” from orbit. That profile highlights cracks or damage. In one study the team applied it to Italy’s Morandi Bridge before it collapsed, SAR-based vibration maps showed unusual energy spikes right at the failing pylon. They even imaged deep tunnels. The HarmonicSAR site reports they “detected for the first time the Gran-Sasso Physics Laboratory at 1.4 km below the Earth using SAR”. In other words, their tomography saw a known underground lab 1400 m under Italy! They’ve also done scans of mountain tunnels (San Gottardo). Biondi was co-author on a 2016 Scientific Reports paper tracking Iraq’s Mosul Dam instability via SAR. That study used spaceborne radar to measure tiny ground motions around the dam over time. It shows that SAR micro-motion techniques can monitor slow structural shifts on a large engineering project. In short, SAR Doppler Tomography isn’t ordinary radar, it’s like using satellites and the Earth’s own background hum to “see” underground. Think of it as applying a CT-scan or ultrasound-like method from orbit. It’s unconventional, but it’s patent-backed and has some peer-reviewed results. RECIEPTS: Synthetic Aperture Radar Doppler Tomography Reveals Details of Undiscovered High-Resolution Internal Structure of the Great Pyramid of Giza Filippo Biondi's SAR/Doppler Patent Perspectives on the Structural Health Monitoring of Bridges by Synthetic Aperture Radar HarmonicSAR (Filippos Website) Mosul Dam SAR/Doppler Project
4:02

🚨They CANNOT SCAN kilometers into SOLID ROCK! That is what skeptics of the Khafre Project’s findings say, and they are correct to point out that Synthetic Aperture Radar (SAR) can't do this. They are also TOTALLY MISTAKEN, here is why: The Khafre team are not using conventional SAR imaging in the way critics assume. They are using a patented form of SAR Doppler Tomography, pioneered by Prof. Filippo Biondi. This is not simple surface imaging. It is a phase-coherent interferometric method that detects subtle Doppler frequency shifts caused by internal micro-vibrations within dense structures. Instead of trying to penetrate rock, it “listens” to tiny seismic vibrations in the stone. Biondi’s trick is to capture micro-motions. Tiny seismic or structural tremors slightly shift the radar’s frequency (Doppler effect). By analyzing these Doppler shifts across multiple SAR images, they can reconstruct a 3D tomographic image of what’s inside, like a CT-scan from space. Prof. Filippop Biondi's patent (PCT/EP2023/064345) explicitly describes processing “coherent vibrational Doppler information” in SAR to allow penetrating 3D imaging "over a depth of several kilometers". In other words, it effectively turns the radar into a spaceborne sonar, using Earth’s natural vibrations to “sound” the subsurface, something ordinary SAR can’t do. A peer-reviewed Remote Sensing paper describes using COSMO-SkyMed SAR data to map new shafts and chambers inside Khufu . This case study in a scientific journal shows the technique in action (with high-res 3D results!). Beyond pyramids, the technique has practical uses. For bridges and infrastructure, Biondi’s SAR Doppler method can extract a structure’s “vibration profile” from orbit. That profile highlights cracks or damage. In one study the team applied it to Italy’s Morandi Bridge before it collapsed, SAR-based vibration maps showed unusual energy spikes right at the failing pylon. They even imaged deep tunnels. The HarmonicSAR site reports they “detected for the first time the Gran-Sasso Physics Laboratory at 1.4 km below the Earth using SAR”. In other words, their tomography saw a known underground lab 1400 m under Italy! They’ve also done scans of mountain tunnels (San Gottardo). Biondi was co-author on a 2016 Scientific Reports paper tracking Iraq’s Mosul Dam instability via SAR. That study used spaceborne radar to measure tiny ground motions around the dam over time. It shows that SAR micro-motion techniques can monitor slow structural shifts on a large engineering project. In short, SAR Doppler Tomography isn’t ordinary radar, it’s like using satellites and the Earth’s own background hum to “see” underground. Think of it as applying a CT-scan or ultrasound-like method from orbit. It’s unconventional, but it’s patent-backed and has some peer-reviewed results. RECIEPTS: Synthetic Aperture Radar Doppler Tomography Reveals Details of Undiscovered High-Resolution Internal Structure of the Great Pyramid of Giza Filippo Biondi's SAR/Doppler Patent Perspectives on the Structural Health Monitoring of Bridges by Synthetic Aperture Radar HarmonicSAR (Filippos Website) Mosul Dam SAR/Doppler Project

Jay Anderson

169,956 Aufrufe • vor 9 Monaten

The power of the Claw, in the palm of a robot hand. Agentic robotics is here! Today, we open-source CaP-X: vibe agents, alive in the physical world. They incarnate as robot arms and humanoids with a rich set of perception APIs, actuation APIs, and auto synthesize skill libraries as they go. CaP-X is a strict superset of our old stack, because policies like VLAs are “just” API calls as well. It solves many tasks zero-shot that a learned policy would struggle with. And we are doing much more than vibing. CaP-X is our most systematic, scientific study on agentic robotics so far: - We build a comprehensive agentic toolkit: perception (SAM3 segmentation, Molmo pointing, depth, point cloud), control (IK solvers, grasp planner, navigation), and visualization (EEF, mask overlays) that work across different robots. - CaP-Gym: LLM’s first Physical Exam! 187 manipulation tasks across RoboSuite, LIBERO-PRO, and BEHAVIOR. Tabletop, bimanual, mobile manipulation. Sim and real. Can’t wait to see the gradients flow from CaP-Gym to the next wave of frontier LLM releases. - CaP-Bench: we benchmark 12 frontier LLMs/VLMs (Gemini, GPT, Opus, Qwen, DeepSeek, Kimi, and more) across 8 evaluation tiers. We systematically vary API abstraction level, agentic harness, and visual grounding methods. Lots of insights in our paper. - CaP-Agent0: a training-free agentic harness that matches or exceeds human expert code on 4 out of 7 tasks without task-specific tuning. - CaP-RL: if you get a gym, you get RL ;). A 7B OSS model jumps from 20% to 72% success after only 50 training iterations. The synthesized programs transfer to real robots with minimal sim-to-real gap. 3 years ago, our team created Voyager, one of the earliest agentic AI that plays and learns in Minecraft continuously. Its key ideas — skill libraries, self-reflection loops, and in-context planning — have since influenced many modern agentic designs. Today, the agent graduates from Minecraft and gets a real job. It’s April Fool’s, but this Claw is getting its hands dirty for real! Link in thread:
1:31

The power of the Claw, in the palm of a robot hand. Agentic robotics is here! Today, we open-source CaP-X: vibe agents, alive in the physical world. They incarnate as robot arms and humanoids with a rich set of perception APIs, actuation APIs, and auto synthesize skill libraries as they go. CaP-X is a strict superset of our old stack, because policies like VLAs are “just” API calls as well. It solves many tasks zero-shot that a learned policy would struggle with. And we are doing much more than vibing. CaP-X is our most systematic, scientific study on agentic robotics so far: - We build a comprehensive agentic toolkit: perception (SAM3 segmentation, Molmo pointing, depth, point cloud), control (IK solvers, grasp planner, navigation), and visualization (EEF, mask overlays) that work across different robots. - CaP-Gym: LLM’s first Physical Exam! 187 manipulation tasks across RoboSuite, LIBERO-PRO, and BEHAVIOR. Tabletop, bimanual, mobile manipulation. Sim and real. Can’t wait to see the gradients flow from CaP-Gym to the next wave of frontier LLM releases. - CaP-Bench: we benchmark 12 frontier LLMs/VLMs (Gemini, GPT, Opus, Qwen, DeepSeek, Kimi, and more) across 8 evaluation tiers. We systematically vary API abstraction level, agentic harness, and visual grounding methods. Lots of insights in our paper. - CaP-Agent0: a training-free agentic harness that matches or exceeds human expert code on 4 out of 7 tasks without task-specific tuning. - CaP-RL: if you get a gym, you get RL ;). A 7B OSS model jumps from 20% to 72% success after only 50 training iterations. The synthesized programs transfer to real robots with minimal sim-to-real gap. 3 years ago, our team created Voyager, one of the earliest agentic AI that plays and learns in Minecraft continuously. Its key ideas — skill libraries, self-reflection loops, and in-context planning — have since influenced many modern agentic designs. Today, the agent graduates from Minecraft and gets a real job. It’s April Fool’s, but this Claw is getting its hands dirty for real! Link in thread:

Jim Fan

81,638 Aufrufe • vor 5 Monaten

What seemed like an intractable problem is now possible: To design proteins with a specified nonlinear mechanical response, capturing complex folding and unfolding mechanisms in singe and few-shot computations. We present ForceGen, an end-to-end algorithm for de novo protein generation based on nonlinear mechanical unfolding responses. Rooted in the physics of protein mechanics, this generative strategy provides a powerful way to design new proteins rapidly, including exquisite and rapid predictions about their dynamical behavior. Proteins, like any other mechanical object, respond to forces in peculiar ways. Think of the different response you'd get from pulling on a steel cable versus pulling on a rubber band, or the difference between honey and glass. Now, we can design proteins with a set of desirable mechanical characteristics, with applications from health to sustainable plastics. The key to solving this problem was to integrate a protein language model with denoising diffusion methods, and using accurate atomistic-level physical simulation data to endow the model a first-principles understanding. ForceGen can solve both forward and inverse tasks: In the forward task, we can predict how stable a protein is, how it will unfold and what the forces involved are, all given just the sequence of amino acids. In the inverse task, we can design new proteins that meet complex nonlinear mechanical signature targets. Read the paper, led by LAMM@MIT postdoc Bo Ni, published in Science Advances: Why do we care about the mechanics of proteins? The mechanics of proteins are critical elements of many living systems - as evidenced in many studies of mechanobiology. Through evolution, nature has presented a set of remarkable protein materials with unique mechanical functions like elastins, silks, keratins or collagens that play crucial roles in biology. However, going beyond natural designs to discover proteins that meet specified mechanical properties remains challenging. So far, the only way to do this was to use existing evolutionary concepts or to manually alter proteins. With our new generative model we can directly design proteins to meet complex nonlinear mechanical property-design objectives. ForceGen leverages deep knowledge on protein sequences from a pretrained protein language model and maps mechanical unfolding responses to create proteins. Via full-atom molecular simulations for direct validation from physical and chemical principles, we demonstrate that the designed proteins are de novo, and fulfill the targeted mechanical properties, including unfolding energy and mechanical strength, and a detailed unfolding force-separation curves. ForceGen offers rapid pathways to explore the enormous mechanobiological protein sequence space unconstrained by biological synthesis, to enable the discovery of new protein materials with superior mechanical properties. B. Ni, D.L. Kaplan, M.J. Buehler, ForceGen: End-to-end de novo protein generation based on nonlinear mechanical unfolding responses using a language diffusion model. Sci. Adv. 10, eadl4000 (2024). DOI: 10.1126/sciadv.adl4000 Codes and model weights available Hugging Face: David Kaplan
0:33

What seemed like an intractable problem is now possible: To design proteins with a specified nonlinear mechanical response, capturing complex folding and unfolding mechanisms in singe and few-shot computations. We present ForceGen, an end-to-end algorithm for de novo protein generation based on nonlinear mechanical unfolding responses. Rooted in the physics of protein mechanics, this generative strategy provides a powerful way to design new proteins rapidly, including exquisite and rapid predictions about their dynamical behavior. Proteins, like any other mechanical object, respond to forces in peculiar ways. Think of the different response you'd get from pulling on a steel cable versus pulling on a rubber band, or the difference between honey and glass. Now, we can design proteins with a set of desirable mechanical characteristics, with applications from health to sustainable plastics. The key to solving this problem was to integrate a protein language model with denoising diffusion methods, and using accurate atomistic-level physical simulation data to endow the model a first-principles understanding. ForceGen can solve both forward and inverse tasks: In the forward task, we can predict how stable a protein is, how it will unfold and what the forces involved are, all given just the sequence of amino acids. In the inverse task, we can design new proteins that meet complex nonlinear mechanical signature targets. Read the paper, led by LAMM@MIT postdoc Bo Ni, published in Science Advances: Why do we care about the mechanics of proteins? The mechanics of proteins are critical elements of many living systems - as evidenced in many studies of mechanobiology. Through evolution, nature has presented a set of remarkable protein materials with unique mechanical functions like elastins, silks, keratins or collagens that play crucial roles in biology. However, going beyond natural designs to discover proteins that meet specified mechanical properties remains challenging. So far, the only way to do this was to use existing evolutionary concepts or to manually alter proteins. With our new generative model we can directly design proteins to meet complex nonlinear mechanical property-design objectives. ForceGen leverages deep knowledge on protein sequences from a pretrained protein language model and maps mechanical unfolding responses to create proteins. Via full-atom molecular simulations for direct validation from physical and chemical principles, we demonstrate that the designed proteins are de novo, and fulfill the targeted mechanical properties, including unfolding energy and mechanical strength, and a detailed unfolding force-separation curves. ForceGen offers rapid pathways to explore the enormous mechanobiological protein sequence space unconstrained by biological synthesis, to enable the discovery of new protein materials with superior mechanical properties. B. Ni, D.L. Kaplan, M.J. Buehler, ForceGen: End-to-end de novo protein generation based on nonlinear mechanical unfolding responses using a language diffusion model. Sci. Adv. 10, eadl4000 (2024). DOI: 10.1126/sciadv.adl4000 Codes and model weights available Hugging Face: David Kaplan

Markus J. Buehler

47,269 Aufrufe • vor 2 Jahren

🚨BREAKING: New Radar Scan Reveals a Massive Engineered Substructure That Looks Like An Energy Grid Beneath the Giza Plateau🚨 Radar engineer Filippo Biondi just dropped the most explosive finding ever reported at Giza: eight clearly man-made, tube-like structures plunging more than a kilometer beneath the Khafre Pyramid and ending in huge 80-meter chambers. The structures are obviously artificially engineered and the synthetic aperture radar Doppler-tomography technique he used has precedent in accurately predicting underground structures (in both commercial and defense use cases). The Egyptian ministry of culture is extremely afraid this finding might rewrite their history. 1) The Core Structure: Eight hollow tubes (with two symmetrical sets of four) sit directly beneath the Khafre Pyramid’s base and run straight down over 1 km into the bedrock. They terminate in a massive chamber roughly 80 meters across. The shape is engineered; nothing in geology produces structures like this. 2) The Method Used: Regular SAR can’t see through rock. Biondi uses a Doppler-tomographic approach: the satellites measure tiny vibrations on the surface, and the inversion reconstructs what’s below from how those vibrations modulate the radar signal. It’s physics, not AI. 3) Independent Replication: The same underground structures appear in data from: • Umbra • Capella Space • ISI • COSMO-SkyMed If this were a glitch or artifact, it wouldn’t repeat across four separate systems. 4) Real-World Validation: Biondi’s method has already been tested against real sites where we know the exact layout. It has: • Mapped the Gran Sasso underground lab with exact accuracy • Reproduced the Osiris Shaft down to ~37 meters • Imaged magma and voids used in active civil-protection monitoring These aren’t speculative models...they match real measurements. 5) Giza As A Unified System: After Khafre, the team scanned the rest of the plateau. Similar tube-like structures miraculously appear beneath Menkaure (a smaller 2+2 pattern) and a single descending tube under the Sphinx. The evidence points to a giant connected system beneath all three monuments. 6) The Tunnel Network: The tomography shows a dense web of tunnels running between the pyramids and toward the Sphinx. Several known surface shafts, now blocked or filled with debris, look like the original access points into this network. 7) Water As A Key Variable: The Osiris Shaft contains water at about 33–37 meters. Biondi thinks water flow is part of how the system operates, possibly tied to vibrational or informational dynamics. He cites Preparata and Del Giudice’s work on coherent water domains but avoids any “power plant” jump. 8) Academia: Biondi already has a peer-reviewed SAR/Doppler tomography paper on the Khufu Pyramid in Remote Sensing. His larger Khafre + plateau paper is now in peer review. The work sticks to hard measurements: geometry, depth, replication. 9) Next Steps: The CAF Project is preparing a proposal to: • Clear out the sealed shafts between Khafre and the Sphinx • Run direct seismic surveys to confirm the satellite data • Enter the tunnel system if Egypt authorizes it At this point, approval from Cairo is the only barrier to verifying what the scans show. If those shafts are opened, the world may be looking at a multi-kilometer engineered complex beneath Giza. Full Documentary 👇
1:07

🚨BREAKING: New Radar Scan Reveals a Massive Engineered Substructure That Looks Like An Energy Grid Beneath the Giza Plateau🚨 Radar engineer Filippo Biondi just dropped the most explosive finding ever reported at Giza: eight clearly man-made, tube-like structures plunging more than a kilometer beneath the Khafre Pyramid and ending in huge 80-meter chambers. The structures are obviously artificially engineered and the synthetic aperture radar Doppler-tomography technique he used has precedent in accurately predicting underground structures (in both commercial and defense use cases). The Egyptian ministry of culture is extremely afraid this finding might rewrite their history. 1) The Core Structure: Eight hollow tubes (with two symmetrical sets of four) sit directly beneath the Khafre Pyramid’s base and run straight down over 1 km into the bedrock. They terminate in a massive chamber roughly 80 meters across. The shape is engineered; nothing in geology produces structures like this. 2) The Method Used: Regular SAR can’t see through rock. Biondi uses a Doppler-tomographic approach: the satellites measure tiny vibrations on the surface, and the inversion reconstructs what’s below from how those vibrations modulate the radar signal. It’s physics, not AI. 3) Independent Replication: The same underground structures appear in data from: • Umbra • Capella Space • ISI • COSMO-SkyMed If this were a glitch or artifact, it wouldn’t repeat across four separate systems. 4) Real-World Validation: Biondi’s method has already been tested against real sites where we know the exact layout. It has: • Mapped the Gran Sasso underground lab with exact accuracy • Reproduced the Osiris Shaft down to ~37 meters • Imaged magma and voids used in active civil-protection monitoring These aren’t speculative models...they match real measurements. 5) Giza As A Unified System: After Khafre, the team scanned the rest of the plateau. Similar tube-like structures miraculously appear beneath Menkaure (a smaller 2+2 pattern) and a single descending tube under the Sphinx. The evidence points to a giant connected system beneath all three monuments. 6) The Tunnel Network: The tomography shows a dense web of tunnels running between the pyramids and toward the Sphinx. Several known surface shafts, now blocked or filled with debris, look like the original access points into this network. 7) Water As A Key Variable: The Osiris Shaft contains water at about 33–37 meters. Biondi thinks water flow is part of how the system operates, possibly tied to vibrational or informational dynamics. He cites Preparata and Del Giudice’s work on coherent water domains but avoids any “power plant” jump. 8) Academia: Biondi already has a peer-reviewed SAR/Doppler tomography paper on the Khufu Pyramid in Remote Sensing. His larger Khafre + plateau paper is now in peer review. The work sticks to hard measurements: geometry, depth, replication. 9) Next Steps: The CAF Project is preparing a proposal to: • Clear out the sealed shafts between Khafre and the Sphinx • Run direct seismic surveys to confirm the satellite data • Enter the tunnel system if Egypt authorizes it At this point, approval from Cairo is the only barrier to verifying what the scans show. If those shafts are opened, the world may be looking at a multi-kilometer engineered complex beneath Giza. Full Documentary 👇

Jesse Michels

891,066 Aufrufe • vor 9 Monaten

Stop Gambling, Start Engineering: The Ultimate Guide To CCXT Algorithmic Trading most traders are essentially walking into a high stakes casino with a blindfold on while the house has a high speed laser aimed directly at their bankroll. if you have ever felt the soul crushing weight of a liquidation notification at three in the morning then you know the market is a 24/7 beast that eats human emotion for breakfast there is a hidden bridge that connects your laptop to almost every major crypto exchange in existence and once you cross it the game changes forever. my name is moon dev i believe that code is the great equalizer because through losing money with liquidations and over trading i knew i had to automate my trading so i learned to code as in the past i spent hundreds of thousands on devs for app, thinking i would not be able to code myself w/ bots you must iterate to success so i decided to learn live on youtube, and now we are here, fully automated systems trading for me instead of getting liquidated. the secret weapon behind this transition is a library called ccxt which acts as a universal translator for exchanges like binance, bybit, and kucoin most people think they need to spend years studying computer science just to place a single trade via code but that is a lie designed to keep you on the sidelines. the reality is that once you understand how to initialize a connection you can control your entire portfolio with just a few lines of logic. it starts with importing the library and setting up your credentials in a way that doesn't leave your keys exposed to the world the first mistake that bankrupts most manual traders is the inability to act fast enough when the trend shifts. when you build a bot the first thing you need to master is the market order because it allows you to enter or exit a position instantly regardless of the price. it is the ultimate panic button for when a strategy goes south or a massive opportunity presents itself while market orders are great for speed they are the fastest way to get eaten alive by fees if you are not careful. this is where the limit order comes into play allowing you to dictate exactly what price you are willing to pay for an asset. by using a create limit order function you can place your bids and asks in the order book and wait for the market to come to you most traders forget that once an order is placed it stays active until it is either filled or manually removed. i have seen countless accounts go to zero because a bot kept piling on buy orders without ever checking to see if the previous ones were canceled. the cancel all orders function is the invisible shield that prevents your algorithm from accidentally over leveraging your account the real magic happens when you realize you can cancel more than just basic limit orders. there are untriggered conditional orders like stop losses and take profits that often hide in the background of an exchange waiting to ruin your day. by passing specific parameters into your cancel function you can wipe the slate clean and ensure your bot is starting from a neutral state every single time if you want to know what the whales are doing before it shows up on a candle chart then you need to be looking at the raw order book. fetching the order book gives you a direct view of every single bid and ask currently sitting on the exchange. this is the most honest data you can get because it represents real money waiting to be filled at specific price levels you can actually parse this data to find the exact top of the bid and the bottom of the ask to ensure your bot always gets the best possible entry. most retail traders are looking at delayed charts while your bot is reading the tape in real time and calculating the spread. this allows you to place orders that are optimized for the current liquidity rather than just guessing where the price might go one of the biggest hurdles in automation is managing the sheer volume of data that an exchange throws at you. when you fetch open high low close volume data you are getting the historical heartbeat of an asset across any timeframe you choose. this data is the foundation of every technical indicator from simple moving averages to complex machine learning models the problem is that raw data is often a mess of lists and dictionaries that are impossible for a human or a simple script to read efficiently. this is why we use pandas to convert that garbage into a structured data frame that looks exactly like a clean spreadsheet. once your data is in a data frame you can calculate rsi or macd with a single line of code and visualize the entire market structure the path to becoming a successful automated trader is not a sprint but a series of iterations toward a system that works. i chose to learn this live in front of the world because i wanted to prove that anyone can escape the cycle of over trading. you don't need a million dollars to start but you do need a system that removes the human element from the equation if you are still clicking buttons on a website then you are competing against machines that can process thousands of data points per second. it is time to stop playing a rigged game and start building your own edge in the market. the code is there for anyone to grab and the only thing standing between you and a fully automated portfolio is the willingness to sit down and write the first line every algorithm you build is a brick in a wall that protects your capital from the emotional swings of the crypto market. i spend my days refining these systems and sharing the process because i know how lonely it feels to lose everything to a flash crash. we are building a community where code is the tool and financial freedom is the goal the final step is realizing that your balance is just a number that your bot needs to manage with cold logic. by fetching your balance frequently your bot can calculate position sizes based on your total equity ensuring that no single trade can ever wipe you out. this is the difference between gambling and systematic trading and it is accessible to anyone with an internet connection i hope you take these tools and start building something that allows you to sleep peacefully while the markets do their thing. the industry is secretive for a reason but we are breaking those walls down one line of code at a time. the journey is long but the reward of never having to worry about a liquidation again is worth every second of the struggle
1:09:25

Stop Gambling, Start Engineering: The Ultimate Guide To CCXT Algorithmic Trading most traders are essentially walking into a high stakes casino with a blindfold on while the house has a high speed laser aimed directly at their bankroll. if you have ever felt the soul crushing weight of a liquidation notification at three in the morning then you know the market is a 24/7 beast that eats human emotion for breakfast there is a hidden bridge that connects your laptop to almost every major crypto exchange in existence and once you cross it the game changes forever. my name is moon dev i believe that code is the great equalizer because through losing money with liquidations and over trading i knew i had to automate my trading so i learned to code as in the past i spent hundreds of thousands on devs for app, thinking i would not be able to code myself w/ bots you must iterate to success so i decided to learn live on youtube, and now we are here, fully automated systems trading for me instead of getting liquidated. the secret weapon behind this transition is a library called ccxt which acts as a universal translator for exchanges like binance, bybit, and kucoin most people think they need to spend years studying computer science just to place a single trade via code but that is a lie designed to keep you on the sidelines. the reality is that once you understand how to initialize a connection you can control your entire portfolio with just a few lines of logic. it starts with importing the library and setting up your credentials in a way that doesn't leave your keys exposed to the world the first mistake that bankrupts most manual traders is the inability to act fast enough when the trend shifts. when you build a bot the first thing you need to master is the market order because it allows you to enter or exit a position instantly regardless of the price. it is the ultimate panic button for when a strategy goes south or a massive opportunity presents itself while market orders are great for speed they are the fastest way to get eaten alive by fees if you are not careful. this is where the limit order comes into play allowing you to dictate exactly what price you are willing to pay for an asset. by using a create limit order function you can place your bids and asks in the order book and wait for the market to come to you most traders forget that once an order is placed it stays active until it is either filled or manually removed. i have seen countless accounts go to zero because a bot kept piling on buy orders without ever checking to see if the previous ones were canceled. the cancel all orders function is the invisible shield that prevents your algorithm from accidentally over leveraging your account the real magic happens when you realize you can cancel more than just basic limit orders. there are untriggered conditional orders like stop losses and take profits that often hide in the background of an exchange waiting to ruin your day. by passing specific parameters into your cancel function you can wipe the slate clean and ensure your bot is starting from a neutral state every single time if you want to know what the whales are doing before it shows up on a candle chart then you need to be looking at the raw order book. fetching the order book gives you a direct view of every single bid and ask currently sitting on the exchange. this is the most honest data you can get because it represents real money waiting to be filled at specific price levels you can actually parse this data to find the exact top of the bid and the bottom of the ask to ensure your bot always gets the best possible entry. most retail traders are looking at delayed charts while your bot is reading the tape in real time and calculating the spread. this allows you to place orders that are optimized for the current liquidity rather than just guessing where the price might go one of the biggest hurdles in automation is managing the sheer volume of data that an exchange throws at you. when you fetch open high low close volume data you are getting the historical heartbeat of an asset across any timeframe you choose. this data is the foundation of every technical indicator from simple moving averages to complex machine learning models the problem is that raw data is often a mess of lists and dictionaries that are impossible for a human or a simple script to read efficiently. this is why we use pandas to convert that garbage into a structured data frame that looks exactly like a clean spreadsheet. once your data is in a data frame you can calculate rsi or macd with a single line of code and visualize the entire market structure the path to becoming a successful automated trader is not a sprint but a series of iterations toward a system that works. i chose to learn this live in front of the world because i wanted to prove that anyone can escape the cycle of over trading. you don't need a million dollars to start but you do need a system that removes the human element from the equation if you are still clicking buttons on a website then you are competing against machines that can process thousands of data points per second. it is time to stop playing a rigged game and start building your own edge in the market. the code is there for anyone to grab and the only thing standing between you and a fully automated portfolio is the willingness to sit down and write the first line every algorithm you build is a brick in a wall that protects your capital from the emotional swings of the crypto market. i spend my days refining these systems and sharing the process because i know how lonely it feels to lose everything to a flash crash. we are building a community where code is the tool and financial freedom is the goal the final step is realizing that your balance is just a number that your bot needs to manage with cold logic. by fetching your balance frequently your bot can calculate position sizes based on your total equity ensuring that no single trade can ever wipe you out. this is the difference between gambling and systematic trading and it is accessible to anyone with an internet connection i hope you take these tools and start building something that allows you to sleep peacefully while the markets do their thing. the industry is secretive for a reason but we are breaking those walls down one line of code at a time. the journey is long but the reward of never having to worry about a liquidation again is worth every second of the struggle

Moon Dev

14,105 Aufrufe • vor 7 Monaten

🧬 BREAKING: Our CRISPR-GPT paper is out TODAY in Nature Biomedical Engineering Nature Biomedical Engineering ! 🤯 We built an AI agent that turns ANYONE into a gene-editing expert in 1 DAY instead of months. An undergrad with ZERO experience achieved 90%+ editing efficiency on their FIRST attempt. 🧵 Here's how we're building expert AI agents for cutting-edge biotechnology: 🎯 The Problem: CRISPR is revolutionary but requires PhD-level expertise, it can take weeks to learn, adopt, and design, analyze a CRISPR experiment for R&D or making life-saving medicine. Even Pro scientists can make small mistakes (e.g. typos in guideRNA or cloning design) that cost months to find out, slowing us down. 💡 Our Solution: CRISPR-GPT - an AI co-pilot from Stanford University Princeton University Google DeepMind that guides you through EVERY step via simple conversation 🔬 Real Results: -Novice researcher: ~90% editing on 1st go -Training time: Months → 1 day -100% success rate in our trials -Even experts save days/weeks on data analysis & troubleshooting 🤖 How it works: Our multi-agent system handles: CRISPR system and delivery method selection, guideRNA design, Protocol generation, Real-time troubleshooting, Data analysis, and beyond. All through natural language! No coding, no complex software. 📊 We benchmarked it extensively: -288 evaluation scenarios/cases -Outperformed GPT-4o on ALL gene editing tasks -Trained on 11 years of expert discussions -Covers knockout, base-editing, prime-editing & epigenetic editing 🌍 Why this matters: -Every lab can now use CRISPR with an AI system distilling expert knowledge and skills. -Every student can learn faster. -Every researcher can tackle bigger challenges without worrying about small mistakes. -Customized CRISPR design can be automated based on your need and the context of R&D workflow. -Agentic AI ensure safety, privacy, and responsibility -We're not just automating gene editing - we're using AI to power scientists to cure diseases. 🚀 Try it yourself! Beta access available at: Paper: Code: Benchmark (companion work, Genome-bench): Co-first and key authors: Yuanhao Qu Kaixuan Huang Ming Yin PIs: Le Cong@Stanford, AI+Bio+Gene-Editing Mengdi Wang Key collaborators: Russ Altman Denny Zhou The future of biology and science is conversational. The future is now. Nature Biomedical Engineering Nature Portfolio #CRISPR #AI #GeneEditing #Biotech #Science #AISafety
1:43

🧬 BREAKING: Our CRISPR-GPT paper is out TODAY in Nature Biomedical Engineering Nature Biomedical Engineering ! 🤯 We built an AI agent that turns ANYONE into a gene-editing expert in 1 DAY instead of months. An undergrad with ZERO experience achieved 90%+ editing efficiency on their FIRST attempt. 🧵 Here's how we're building expert AI agents for cutting-edge biotechnology: 🎯 The Problem: CRISPR is revolutionary but requires PhD-level expertise, it can take weeks to learn, adopt, and design, analyze a CRISPR experiment for R&D or making life-saving medicine. Even Pro scientists can make small mistakes (e.g. typos in guideRNA or cloning design) that cost months to find out, slowing us down. 💡 Our Solution: CRISPR-GPT - an AI co-pilot from Stanford University Princeton University Google DeepMind that guides you through EVERY step via simple conversation 🔬 Real Results: -Novice researcher: ~90% editing on 1st go -Training time: Months → 1 day -100% success rate in our trials -Even experts save days/weeks on data analysis & troubleshooting 🤖 How it works: Our multi-agent system handles: CRISPR system and delivery method selection, guideRNA design, Protocol generation, Real-time troubleshooting, Data analysis, and beyond. All through natural language! No coding, no complex software. 📊 We benchmarked it extensively: -288 evaluation scenarios/cases -Outperformed GPT-4o on ALL gene editing tasks -Trained on 11 years of expert discussions -Covers knockout, base-editing, prime-editing & epigenetic editing 🌍 Why this matters: -Every lab can now use CRISPR with an AI system distilling expert knowledge and skills. -Every student can learn faster. -Every researcher can tackle bigger challenges without worrying about small mistakes. -Customized CRISPR design can be automated based on your need and the context of R&D workflow. -Agentic AI ensure safety, privacy, and responsibility -We're not just automating gene editing - we're using AI to power scientists to cure diseases. 🚀 Try it yourself! Beta access available at: Paper: Code: Benchmark (companion work, Genome-bench): Co-first and key authors: Yuanhao Qu Kaixuan Huang Ming Yin PIs: Le Cong@Stanford, AI+Bio+Gene-Editing Mengdi Wang Key collaborators: Russ Altman Denny Zhou The future of biology and science is conversational. The future is now. Nature Biomedical Engineering Nature Portfolio #CRISPR #AI #GeneEditing #Biotech #Science #AISafety

Le Cong@Stanford, AI+Bio+Gene-Editing

111,894 Aufrufe • vor 1 Jahr

Our new AI model, SparksMatter, discovered CaMg₂Si₂ - a Ca-filled Mg-Si Zintl silicide - as a thermoelectric built only from stable, non-toxic, earth-abundant elements. Thermoelectrics are solid-state materials that convert heat directly into electricity (and electricity into cooling) with no moving parts, which makes them a key technology for harvesting the vast amounts of waste heat from engines, industry and electronics, and even fusion - but today's best ones rely on scarce or toxic elements like tellurium, lead and bismuth. This is why an earth-abundant, non-toxic candidate matters. Our model's physical reasoning to come up with the design: Mg₂Si is a known earth-abundant thermoelectric but conducts heat too well; a heavy, weakly bound Ca cation in the Mg-Si framework should scatter phonons while keeping a moderate band gap. It generated 100 Ca-Mg-Si crystals with MatterGen, kept the six within 0.05 eV/atom of the convex hull (via MatterSim), and predicted band gaps of 0.44-0.57 eV and bulk moduli of 53-54 GPa (CGCNN). Follow-up lattice dynamics found three CaMg₂Si₂ polymorphs dynamically stable, with lattice thermal conductivity ≈6 W m⁻¹ K⁻¹ at 300 K and ≈2 at 1000 K. The AI proposed a chemical hypothesis first, then developed and applied a separate generative/physics pipeline to test it, and six surviving structures came back with that hypothesized composition. The video replays the reasoning process. New paper out with Alireza Ghafarollahi in npj Computational Materials: SparksMatter, an AI that runs the full in-silico inorganic materials discovery cycle - ideation, planning, computational experimentation, critique and reporting - from a single plain-language query. Why this matters: conventional ML models for materials are typically single-shot predictors or generators. They can predict a property or propose a structure, but they do not organize the next scientific step. Discovery, instead, works as a loop: hypothesize, test, critique, revise. The key advance here is a deep reasoning layer that incorporates physics to decide which scientific tool to use, how to interpret the result, and what to change as next step. How it works: SparksMatter spawns a suite of AI agents - scientists, planners, coders, reviewers and critics - that write and execute code against materials tools: Materials Project retrieval; MatterGen for generative crystal design conditioned on chemistry, band gap or bulk modulus; MatterSim for relaxation and convex-hull stability; CGCNN for property prediction. Adversarial agents check each phase, the system revises its ideas, plans and code from execution results, documents its own limitations, and delivers a scientific report with a validation roadmap spanning DFT, phonons, transport, synthesis and characterization. Two more discovery tasks SparksMatter ran autonomously: 1⃣Soft inorganic semiconductors: generated 112 structures conditioned on low stiffness and narrowed them to 59 candidates absent from the Materials Project after toxicity, stability, electronic, mechanical and database screening - bulk moduli 11-24 GPa, band gaps 0.4-3.9 eV. 2⃣Lead-free perovskites: filtered 154,879 Materials Project entries to 162 Pb-free ABO₃ candidates meeting structural, toxicity, stability and band-gap criteria, including LaAlO₃, BaZrO₃, SrSnO₃, CaTiO₃ and SrTiO₃. Benchmark: the same three tasks were given to frontier reasoning models acting as expert materials scientists with web browsing but without the generation and prediction tools. A blinded LLM evaluator scored every response ten times on relevance, scientific soundness, novelty, and depth and rigor. SparksMatter scored highest in aggregate, with its strongest advantages in novelty and depth & rigor. Its main limitation was scientific soundness because much of the core screening still relied on surrogate models rather than direct first-principles or experimental validation - a gap the system identified, documented, and mapped out how to close. Takeaway: putting generative models, executable code and physics-based simulators inside the reasoning loop lets an AI propose structures outside existing databases, test them, reject weak candidates, and say what evidence is still missing.
1:15

Our new AI model, SparksMatter, discovered CaMg₂Si₂ - a Ca-filled Mg-Si Zintl silicide - as a thermoelectric built only from stable, non-toxic, earth-abundant elements. Thermoelectrics are solid-state materials that convert heat directly into electricity (and electricity into cooling) with no moving parts, which makes them a key technology for harvesting the vast amounts of waste heat from engines, industry and electronics, and even fusion - but today's best ones rely on scarce or toxic elements like tellurium, lead and bismuth. This is why an earth-abundant, non-toxic candidate matters. Our model's physical reasoning to come up with the design: Mg₂Si is a known earth-abundant thermoelectric but conducts heat too well; a heavy, weakly bound Ca cation in the Mg-Si framework should scatter phonons while keeping a moderate band gap. It generated 100 Ca-Mg-Si crystals with MatterGen, kept the six within 0.05 eV/atom of the convex hull (via MatterSim), and predicted band gaps of 0.44-0.57 eV and bulk moduli of 53-54 GPa (CGCNN). Follow-up lattice dynamics found three CaMg₂Si₂ polymorphs dynamically stable, with lattice thermal conductivity ≈6 W m⁻¹ K⁻¹ at 300 K and ≈2 at 1000 K. The AI proposed a chemical hypothesis first, then developed and applied a separate generative/physics pipeline to test it, and six surviving structures came back with that hypothesized composition. The video replays the reasoning process. New paper out with Alireza Ghafarollahi in npj Computational Materials: SparksMatter, an AI that runs the full in-silico inorganic materials discovery cycle - ideation, planning, computational experimentation, critique and reporting - from a single plain-language query. Why this matters: conventional ML models for materials are typically single-shot predictors or generators. They can predict a property or propose a structure, but they do not organize the next scientific step. Discovery, instead, works as a loop: hypothesize, test, critique, revise. The key advance here is a deep reasoning layer that incorporates physics to decide which scientific tool to use, how to interpret the result, and what to change as next step. How it works: SparksMatter spawns a suite of AI agents - scientists, planners, coders, reviewers and critics - that write and execute code against materials tools: Materials Project retrieval; MatterGen for generative crystal design conditioned on chemistry, band gap or bulk modulus; MatterSim for relaxation and convex-hull stability; CGCNN for property prediction. Adversarial agents check each phase, the system revises its ideas, plans and code from execution results, documents its own limitations, and delivers a scientific report with a validation roadmap spanning DFT, phonons, transport, synthesis and characterization. Two more discovery tasks SparksMatter ran autonomously: 1⃣Soft inorganic semiconductors: generated 112 structures conditioned on low stiffness and narrowed them to 59 candidates absent from the Materials Project after toxicity, stability, electronic, mechanical and database screening - bulk moduli 11-24 GPa, band gaps 0.4-3.9 eV. 2⃣Lead-free perovskites: filtered 154,879 Materials Project entries to 162 Pb-free ABO₃ candidates meeting structural, toxicity, stability and band-gap criteria, including LaAlO₃, BaZrO₃, SrSnO₃, CaTiO₃ and SrTiO₃. Benchmark: the same three tasks were given to frontier reasoning models acting as expert materials scientists with web browsing but without the generation and prediction tools. A blinded LLM evaluator scored every response ten times on relevance, scientific soundness, novelty, and depth and rigor. SparksMatter scored highest in aggregate, with its strongest advantages in novelty and depth & rigor. Its main limitation was scientific soundness because much of the core screening still relied on surrogate models rather than direct first-principles or experimental validation - a gap the system identified, documented, and mapped out how to close. Takeaway: putting generative models, executable code and physics-based simulators inside the reasoning loop lets an AI propose structures outside existing databases, test them, reject weak candidates, and say what evidence is still missing.

Markus J. Buehler

42,151 Aufrufe • vor 21 Tagen

I Cracked Polymarket Using Claude Opus 4.6: The 96,000 Dollar Script For 5 Minute High Leverage Windows most traders are currently sitting at their desks fighting a losing battle against a digital wall because they do not realize the house always wins against human emotion. while the crowd is busy chasing the next meme coin or getting washed out in a single wick an automated agent just pulled nearly a hundred thousand dollars out of thin air using nothing but raw logic. i have seen people blow their life savings in these five minute windows because they treated a high leverage prediction market like a playground instead of a laboratory i am moon dev and i believe that code is the great equalizer because through losing money with liquidations and over trading i knew i had to automate my trading. in the past i spent hundreds of thousands on devs for apps thinking i would not be able to code myself which was a massive waste of my time and resources. with bots you must iterate to success so i decided to learn live on youtube and now we are here with fully automated systems trading for me instead of getting liquidated by the market the five minute markets on polymarket are essentially a high speed game of musical chairs where the person left standing is usually the one with the fastest script. leverage makes these markets extremely dangerous because it amplifies every mistake you make until your account is completely empty. the only way to survive this environment is to stop trading based on a gut feeling and start trading based on a stress tested mathematical edge the real breakthrough happened when i started using claude opus four point six to write the execution code for these specific five minute windows. having an ai agent that can analyze microstructure data means you can find trends that are completely invisible to the naked eye. it is essentially like having a team of twenty engineers working for you around the clock without the communication lag or the massive overhead costs some of our back tests show a sixty four percent win rate which sounds like a dream to anyone who has ever spent a night staring at a red screen. however the return on these tests varies wildly based on a few specific changes to the strategy parameters and histogram filters. i found that a return of forty one thousand dollars can jump to nearly double that just by adjusting how the bot handles the macd histogram threshold the trap that most people fall into is thinking that a good back test is a license to print money immediately without any further validation. this is a dangerous lie that leads to huge losses because the market in the past is not a perfect mirror of what is going to happen today. that is why i never launch a bot with full size until it has survived the incubation phase where it trades with ten dollars at a time incubation is the ultimate reality check for any trading strategy regardless of how good the numbers look on a computer screen. it is nerve wracking to watch a bot enter its first real trade even if the size is small because that is the moment theory meets reality. most of the bots that pass a back test will fail during the first forty eight hours of incubation and that is exactly why this step is non negotiable the data i use to build these systems covers over two hundred weeks of historical one minute candles to ensure the results are robust and not just luck. we are currently moving toward a machine learning approach where the system can adapt to changing market conditions without me having to intervene. this means the bot will eventually be able to recognize when a high leverage window is too risky and simply wait for a better entry the strategy itself relies heavily on macd variations which is a well known indicator but it is used here with a very specific and proprietary twist. by filtering for trades that hit a specific threshold we can ignore the random price noise that usually liquidates manual traders. we look for an edge of at least six percent which is enough to cover all platform fees and still leave a significant profit on the table i used to think that being a successful trader meant being a genius who could predict the future with a magical crystal ball. the truth is far more boring because success is just about researching an idea and testing it until the data proves it works in the past. then you just let the bot do the work while you go live your life instead of being a slave to the candle sticks and charts this world is changing fast and the people who learn to leverage ai to automate their thinking are going to be the ones who win the next decade. i am not asking you to trust a back test or a screenshot from a website because i want you to trust the process of testing it yourself. code allows you to take your life back from the screens and finally stop the cycle of over trading and emotional liquidations the difference between the traders who make it and the people who blow up is simply the willingness to iterate on their ideas daily. you might fail on your first ten bots but the eleventh one might be the script that changes your entire financial trajectory forever. it is about staying in the game long enough for the math to finally work in your favor and removing the human heart from the execution every day i am back testing and researching new ideas to see if they can survive the stress of real market data. i launch these live bots and let them run for seventy two hours to see if they can handle the pressure of the current market trend. while everyone else is coping and complaining about market volatility we are just adjusting our parameters and letting the ai find the next profitable window vibe coding with claude opus four point six has changed the speed at which i can deploy a new strategy from weeks down to just a few minutes. you can give the ai a general strategy idea and it builds the entire trading infrastructure for you while you focus on the logic. this speed is the ultimate advantage in a market that moves as fast as a five minute prediction window on the blockchain the future of trading is not found in a chat room or a paid signal group but in the code you write and the data you process. i believe that everyone has the ability to become an automated trader if they are willing to put in the work to learn the scripts. it is the only way to escape the trap of the nine to five and the anxiety of manual hand trading in a manipulated market i want you to understand that the ninety six thousand dollar returns i see are the result of hundreds of failed tests that never saw the light of day. you have to be willing to look at a failing bot and kill it without emotion so you can move on to the next research project. that is the quantitative mindset that separates the winners from the people who are just gambling with their savings if you are ready to stop being the liquidity for the big players then it is time to start building your own automated army of bots. for the cost of a few cups of coffee you can get access to the road map and the scripts that are driving these results. i am here every day showing you the process because i want to see more people use code to find their financial freedom and beat the house at its own game
11:04

I Cracked Polymarket Using Claude Opus 4.6: The 96,000 Dollar Script For 5 Minute High Leverage Windows most traders are currently sitting at their desks fighting a losing battle against a digital wall because they do not realize the house always wins against human emotion. while the crowd is busy chasing the next meme coin or getting washed out in a single wick an automated agent just pulled nearly a hundred thousand dollars out of thin air using nothing but raw logic. i have seen people blow their life savings in these five minute windows because they treated a high leverage prediction market like a playground instead of a laboratory i am moon dev and i believe that code is the great equalizer because through losing money with liquidations and over trading i knew i had to automate my trading. in the past i spent hundreds of thousands on devs for apps thinking i would not be able to code myself which was a massive waste of my time and resources. with bots you must iterate to success so i decided to learn live on youtube and now we are here with fully automated systems trading for me instead of getting liquidated by the market the five minute markets on polymarket are essentially a high speed game of musical chairs where the person left standing is usually the one with the fastest script. leverage makes these markets extremely dangerous because it amplifies every mistake you make until your account is completely empty. the only way to survive this environment is to stop trading based on a gut feeling and start trading based on a stress tested mathematical edge the real breakthrough happened when i started using claude opus four point six to write the execution code for these specific five minute windows. having an ai agent that can analyze microstructure data means you can find trends that are completely invisible to the naked eye. it is essentially like having a team of twenty engineers working for you around the clock without the communication lag or the massive overhead costs some of our back tests show a sixty four percent win rate which sounds like a dream to anyone who has ever spent a night staring at a red screen. however the return on these tests varies wildly based on a few specific changes to the strategy parameters and histogram filters. i found that a return of forty one thousand dollars can jump to nearly double that just by adjusting how the bot handles the macd histogram threshold the trap that most people fall into is thinking that a good back test is a license to print money immediately without any further validation. this is a dangerous lie that leads to huge losses because the market in the past is not a perfect mirror of what is going to happen today. that is why i never launch a bot with full size until it has survived the incubation phase where it trades with ten dollars at a time incubation is the ultimate reality check for any trading strategy regardless of how good the numbers look on a computer screen. it is nerve wracking to watch a bot enter its first real trade even if the size is small because that is the moment theory meets reality. most of the bots that pass a back test will fail during the first forty eight hours of incubation and that is exactly why this step is non negotiable the data i use to build these systems covers over two hundred weeks of historical one minute candles to ensure the results are robust and not just luck. we are currently moving toward a machine learning approach where the system can adapt to changing market conditions without me having to intervene. this means the bot will eventually be able to recognize when a high leverage window is too risky and simply wait for a better entry the strategy itself relies heavily on macd variations which is a well known indicator but it is used here with a very specific and proprietary twist. by filtering for trades that hit a specific threshold we can ignore the random price noise that usually liquidates manual traders. we look for an edge of at least six percent which is enough to cover all platform fees and still leave a significant profit on the table i used to think that being a successful trader meant being a genius who could predict the future with a magical crystal ball. the truth is far more boring because success is just about researching an idea and testing it until the data proves it works in the past. then you just let the bot do the work while you go live your life instead of being a slave to the candle sticks and charts this world is changing fast and the people who learn to leverage ai to automate their thinking are going to be the ones who win the next decade. i am not asking you to trust a back test or a screenshot from a website because i want you to trust the process of testing it yourself. code allows you to take your life back from the screens and finally stop the cycle of over trading and emotional liquidations the difference between the traders who make it and the people who blow up is simply the willingness to iterate on their ideas daily. you might fail on your first ten bots but the eleventh one might be the script that changes your entire financial trajectory forever. it is about staying in the game long enough for the math to finally work in your favor and removing the human heart from the execution every day i am back testing and researching new ideas to see if they can survive the stress of real market data. i launch these live bots and let them run for seventy two hours to see if they can handle the pressure of the current market trend. while everyone else is coping and complaining about market volatility we are just adjusting our parameters and letting the ai find the next profitable window vibe coding with claude opus four point six has changed the speed at which i can deploy a new strategy from weeks down to just a few minutes. you can give the ai a general strategy idea and it builds the entire trading infrastructure for you while you focus on the logic. this speed is the ultimate advantage in a market that moves as fast as a five minute prediction window on the blockchain the future of trading is not found in a chat room or a paid signal group but in the code you write and the data you process. i believe that everyone has the ability to become an automated trader if they are willing to put in the work to learn the scripts. it is the only way to escape the trap of the nine to five and the anxiety of manual hand trading in a manipulated market i want you to understand that the ninety six thousand dollar returns i see are the result of hundreds of failed tests that never saw the light of day. you have to be willing to look at a failing bot and kill it without emotion so you can move on to the next research project. that is the quantitative mindset that separates the winners from the people who are just gambling with their savings if you are ready to stop being the liquidity for the big players then it is time to start building your own automated army of bots. for the cost of a few cups of coffee you can get access to the road map and the scripts that are driving these results. i am here every day showing you the process because i want to see more people use code to find their financial freedom and beat the house at its own game

Moon Dev

10,921 Aufrufe • vor 5 Monaten

Thermodynamic computing is here There is a new computing paradigm emerging from the noise, and its arrival may be as significant as the dawn of deep learning or the advent of cloud virtualization. A new company, Extropic, has just launched its first thermodynamic computer, a device they call a TSU, or Thermal Sampling Unit. While the web is already filling with deep technical dives, what’s more important for most of us is building a clear intuition for what this technology is, how it’s fundamentally different from anything that’s come before, and why it’s generating so much excitement. This isn’t just another chip; it’s a new way to think about computation itself. Seeing is Believing: Solving Puzzles in One Shot To understand what a TSU does, let’s look at two classic, notoriously difficult computer science problems: Sudoku and the Eight Queens problem. When you or I solve a Sudoku, we use a process of sequential logic, guess-and-check, and backtracking. We make an assumption, follow its logical conclusion, and if we hit a dead end, we erase and try again. A classical computer does the same, just much faster. A TSU, however, approaches this in a completely different way. Using a TSU simulator, one can “program” the problem by first clamping the known values—the clues already on the board. Then, you program in the constraints: no duplicate numbers in any row, column, or 3x3 square. With the problem thus defined, the TSU doesn’t “search” for a solution; it anneals one. In a single computational step, the solution simply emerges, backfilling all the empty squares correctly. The same principle applies to the Eight Queens problem, a challenge to place eight queens on a chessboard so that none can attack any other. This is a complex combinatorial problem with 92 distinct solutions. A classical computer would have to iteratively search for these. A TSU, by contrast, can be programmed with the constraints (the “anti-affinity” between queens on the same row, column, or diagonal) and then set to sample the “solution space.” In this context, a valid solution is one with a “problem energy” of zero. The TSU’s physical nature allows it to naturally find these zero-energy states. A simulation of this process shows the TSU discovering all 92 unique solutions, demonstrating its ability to not just find an answer, but to explore the entire landscape of all correct answers. This is a fundamentally new approach, one that bypasses the brute-force, iterative methods we’ve relied on for decades. The Physics of Computation: Using Noise, Not Fighting It This new power comes from a radical design philosophy. For the last 70 years, computing has been about one thing: order. We build chips that are deterministic, logical, and precise. The great enemy has always been noise, heat, and randomness. We spend billions on cooling and error correction to eliminate these very things. Quantum computing, in many ways, is the ultimate expression of this, requiring temperatures near absolute zero to eliminate all thermal noise and achieve quantum coherence. Thermodynamic computing is the polar opposite. It doesn’t fight the noise; it uses it. The TSU is built on the understanding that the natural, stochastic noise from “leaky” transistors—the very randomness we’ve tried to engineer out of existence—is itself a powerful computational resource. Think of it this way: a GPU, which is central to today’s AI, has to simulate noise. When a generative AI model creates a new image or sentence, it’s using complex algorithms to fake randomness. The TSU doesn’t need to fake it; it harnesses the actual physical randomness of thermodynamics. It is a piece of hardware that directly computes with probability. This makes it a hybrid, sitting somewhere between a purely analog computer (which might use light or sound waves to compute) and a digital GPU. It’s a physical device that leverages the laws of physics itself to find solutions, rather than just using logic gates to simulate them. From a Lost Hiker to a Million Bouncy Balls Perhaps the best way to build intuition is with a metaphor. Imagine that solving a complex optimization problem is like trying to find the lowest point of altitude in a 100-square-mile mountainous landscape. Classical computing, using an algorithm like gradient descent, is like being a single hiker dropped into this landscape at night. You have no map or satellite view. All you have is an altimeter and the sensation of the slope under your feet. You can only take one step at a time, always walking downhill, hoping you don’t get stuck in a small local valley when the true, lowest canyon is miles away. Thermodynamic computing is a completely different approach. It’s like having a million bouncy balls and a helicopter. You drop all million balls simultaneously across the entire 100-square-mile landscape. Then, you “turn on an earthquake,” shaking the entire system. The balls bounce and jostle, but as the shaking (the “annealing”) subsides, where do they all end up? They naturally settle into the lowest points. The balls that collect in the deepest valley represent the optimal solution. The TSU is, in essence, a physical device for dropping those million balls at once and letting the laws of thermodynamics find the lowest “energy” state for you, all at the same time. Beyond Puzzles: The Real-World Impact This is far more than just a clever way to solve brain teasers. This ability to instantly find the lowest energy state for a complex, constrained system has staggering real-world applications. One of the most immediate is protein folding. Companies like Google’s DeepMind have made incredible progress with AI like AlphaFold, which predicts protein structures. But this is still a predictive model trained on existing data. A TSU could potentially solve the folding problem directly, treating the protein as a system of atomic affinities and repulsions and finding its most stable, lowest-energy configuration almost instantaneously. This could revolutionize drug discovery and materials science. An even more profound possibility lies in nuclear fusion. One of the greatest engineering challenges in history is controlling the superheated plasma within a tokamak reactor. This requires shaping unimaginably complex magnetic containment fields in real-time to prevent the plasma from touching the reactor walls. This is a real-time optimization problem so complex it’s currently beyond our capabilities. A TSU, however, could be fast enough. Its ability to compute with electricity itself, rather than abstracting the problem through layers of software, might allow it to update the magnetic fields fast enough to stabilize the fusion reaction. One could even imagine a future where thermodynamic computing elements are built directly into the tokamak’s walls, allowing the reactor to physically and intelligently react to the plasma’s state in real time. A ‘GPT-2 Moment’ for a New Era It’s easy to become numb to hype, but what we are witnessing with the TSU feels different. This is what you might call a “GPT-2 moment.” For those who were there, GPT-2 was the first generative AI model that wasn’t just a toy; it was the first time you could play with it at home and see the spark of true generative intelligence. It was the precursor that pointed directly to the GPT-3 and ChatGPT revolution that has since changed the world. This TSU has that same feel. It’s the “SDK” for a new computing paradigm. This technology is as different from classical computing as quantum computing is, but with a critical difference: a team of 15 built this in two years, and it runs at room temperature on your desk. Quantum computing has seen decades of work and billions in funding, and it still hasn’t produced a commercially viable, scalable machine. The TSU is here now. Based on a two-decade-long career at the cutting edge of technology—from seeing the obvious future of virtualization in 2007 to an early conviction in deep learning and GPT—this has all the same hallmarks of a fundamental, world-changing shift. We are not just building faster calculators; we are learning to compute with the universe itself. Pay close attention to this. This is the next big thing.
18:01

Thermodynamic computing is here There is a new computing paradigm emerging from the noise, and its arrival may be as significant as the dawn of deep learning or the advent of cloud virtualization. A new company, Extropic, has just launched its first thermodynamic computer, a device they call a TSU, or Thermal Sampling Unit. While the web is already filling with deep technical dives, what’s more important for most of us is building a clear intuition for what this technology is, how it’s fundamentally different from anything that’s come before, and why it’s generating so much excitement. This isn’t just another chip; it’s a new way to think about computation itself. Seeing is Believing: Solving Puzzles in One Shot To understand what a TSU does, let’s look at two classic, notoriously difficult computer science problems: Sudoku and the Eight Queens problem. When you or I solve a Sudoku, we use a process of sequential logic, guess-and-check, and backtracking. We make an assumption, follow its logical conclusion, and if we hit a dead end, we erase and try again. A classical computer does the same, just much faster. A TSU, however, approaches this in a completely different way. Using a TSU simulator, one can “program” the problem by first clamping the known values—the clues already on the board. Then, you program in the constraints: no duplicate numbers in any row, column, or 3x3 square. With the problem thus defined, the TSU doesn’t “search” for a solution; it anneals one. In a single computational step, the solution simply emerges, backfilling all the empty squares correctly. The same principle applies to the Eight Queens problem, a challenge to place eight queens on a chessboard so that none can attack any other. This is a complex combinatorial problem with 92 distinct solutions. A classical computer would have to iteratively search for these. A TSU, by contrast, can be programmed with the constraints (the “anti-affinity” between queens on the same row, column, or diagonal) and then set to sample the “solution space.” In this context, a valid solution is one with a “problem energy” of zero. The TSU’s physical nature allows it to naturally find these zero-energy states. A simulation of this process shows the TSU discovering all 92 unique solutions, demonstrating its ability to not just find an answer, but to explore the entire landscape of all correct answers. This is a fundamentally new approach, one that bypasses the brute-force, iterative methods we’ve relied on for decades. The Physics of Computation: Using Noise, Not Fighting It This new power comes from a radical design philosophy. For the last 70 years, computing has been about one thing: order. We build chips that are deterministic, logical, and precise. The great enemy has always been noise, heat, and randomness. We spend billions on cooling and error correction to eliminate these very things. Quantum computing, in many ways, is the ultimate expression of this, requiring temperatures near absolute zero to eliminate all thermal noise and achieve quantum coherence. Thermodynamic computing is the polar opposite. It doesn’t fight the noise; it uses it. The TSU is built on the understanding that the natural, stochastic noise from “leaky” transistors—the very randomness we’ve tried to engineer out of existence—is itself a powerful computational resource. Think of it this way: a GPU, which is central to today’s AI, has to simulate noise. When a generative AI model creates a new image or sentence, it’s using complex algorithms to fake randomness. The TSU doesn’t need to fake it; it harnesses the actual physical randomness of thermodynamics. It is a piece of hardware that directly computes with probability. This makes it a hybrid, sitting somewhere between a purely analog computer (which might use light or sound waves to compute) and a digital GPU. It’s a physical device that leverages the laws of physics itself to find solutions, rather than just using logic gates to simulate them. From a Lost Hiker to a Million Bouncy Balls Perhaps the best way to build intuition is with a metaphor. Imagine that solving a complex optimization problem is like trying to find the lowest point of altitude in a 100-square-mile mountainous landscape. Classical computing, using an algorithm like gradient descent, is like being a single hiker dropped into this landscape at night. You have no map or satellite view. All you have is an altimeter and the sensation of the slope under your feet. You can only take one step at a time, always walking downhill, hoping you don’t get stuck in a small local valley when the true, lowest canyon is miles away. Thermodynamic computing is a completely different approach. It’s like having a million bouncy balls and a helicopter. You drop all million balls simultaneously across the entire 100-square-mile landscape. Then, you “turn on an earthquake,” shaking the entire system. The balls bounce and jostle, but as the shaking (the “annealing”) subsides, where do they all end up? They naturally settle into the lowest points. The balls that collect in the deepest valley represent the optimal solution. The TSU is, in essence, a physical device for dropping those million balls at once and letting the laws of thermodynamics find the lowest “energy” state for you, all at the same time. Beyond Puzzles: The Real-World Impact This is far more than just a clever way to solve brain teasers. This ability to instantly find the lowest energy state for a complex, constrained system has staggering real-world applications. One of the most immediate is protein folding. Companies like Google’s DeepMind have made incredible progress with AI like AlphaFold, which predicts protein structures. But this is still a predictive model trained on existing data. A TSU could potentially solve the folding problem directly, treating the protein as a system of atomic affinities and repulsions and finding its most stable, lowest-energy configuration almost instantaneously. This could revolutionize drug discovery and materials science. An even more profound possibility lies in nuclear fusion. One of the greatest engineering challenges in history is controlling the superheated plasma within a tokamak reactor. This requires shaping unimaginably complex magnetic containment fields in real-time to prevent the plasma from touching the reactor walls. This is a real-time optimization problem so complex it’s currently beyond our capabilities. A TSU, however, could be fast enough. Its ability to compute with electricity itself, rather than abstracting the problem through layers of software, might allow it to update the magnetic fields fast enough to stabilize the fusion reaction. One could even imagine a future where thermodynamic computing elements are built directly into the tokamak’s walls, allowing the reactor to physically and intelligently react to the plasma’s state in real time. A ‘GPT-2 Moment’ for a New Era It’s easy to become numb to hype, but what we are witnessing with the TSU feels different. This is what you might call a “GPT-2 moment.” For those who were there, GPT-2 was the first generative AI model that wasn’t just a toy; it was the first time you could play with it at home and see the spark of true generative intelligence. It was the precursor that pointed directly to the GPT-3 and ChatGPT revolution that has since changed the world. This TSU has that same feel. It’s the “SDK” for a new computing paradigm. This technology is as different from classical computing as quantum computing is, but with a critical difference: a team of 15 built this in two years, and it runs at room temperature on your desk. Quantum computing has seen decades of work and billions in funding, and it still hasn’t produced a commercially viable, scalable machine. The TSU is here now. Based on a two-decade-long career at the cutting edge of technology—from seeing the obvious future of virtualization in 2007 to an early conviction in deep learning and GPT—this has all the same hallmarks of a fundamental, world-changing shift. We are not just building faster calculators; we are learning to compute with the universe itself. Pay close attention to this. This is the next big thing.

David Shapiro (L/0)

83,649 Aufrufe • vor 10 Monaten