Загрузка видео...

Не удалось загрузить видео

Возникла проблема при загрузке этого видео. Это может быть связано с временными проблемами сети или видео может быть недоступно.

На главную

Continuous-video agents (computer use, robotics, static scenes) burn compute re-ingesting pixels that didn't move. VLMaxxing teaches a frozen video VLM to skip the reruns. 54 fps perception on Gemma 4 26B, training-free, no accuracy drift. w/ JF Bastien (arXiv 2605.03351)

Sam D'Amico

33,903 subscribers

63,409 просмотров • 2 месяцев назад •via X (Twitter)

Образование Наука и технологии

Anya Rossi• Live Now

Private livecam show

Комментарии: 0

Нет доступных комментариев

Здесь появятся комментарии из оригинального поста

Похожие видео

How to turn a PowerPoint into a full video with voiceover (without recording anything). Turning slides into a video meant: ↳ Editing timelines, recording audio, fixing pacing, basically rebuilding your deck from scratch. Not anymore… With Synthesia 🎥 you just upload a PPT and get a video: ➟ Your slides become editable video scenes ➟ Your speaker notes become a natural voiceover ➟ An AI avatar delivers your content on-screen ➟ Your videos can be translated into 140+ languages ➟ All while keeping your original layout, fonts, and design intact No camera setup. No retakes. No audio editing. ↳ And this feature is free to use. Just: upload a PPT → AI generate video → download it. This turns your "static" assets into high-value content: ✓ Training deck sitting in your Google Drive? Now a watchable video. ✓ Sales slides nobody reads? Now a 3-minute demo with a presenter. ✓ Onboarding presentation? Now a proper video course. 👉 And the best part: It’s a free feature. 📌 Convert PowerPoints to videos for free: 🔄 Repost to help others turn slides into videos — #AI #Video #PowerPoint #SynthesiaPartnership

Andrew Bolis

41,775 просмотров • 6 месяцев назад

Want a free private AI agent on your laptop? Here’s the simple setup. 👇 1. Install Hermes Agent from GitHub. 2. Install Ollama. 3. Download Gemma 4 26B. 4. Use Docker to keep everything contained. 5. Add Firecrawl so the agent can research the web. 6. Connect Telegram so you can chat with it from your phone. 7. Turn on context compression so it stays sharp. 8. Set the daily reset for 4 AM so it refreshes while you sleep. What you get at the end: → local AI agent → private workflow → web research built in → self improving skills → no monthly model rental That is a serious setup for anyone building with AI. Save this video, you’ll have the easiest beginner roadmap to follow later. Want the SOP? DM me. 💬

Want a free private AI agent on your laptop? Here’s the simple setup. 👇 1. Install Hermes Agent from GitHub. 2. Install Ollama. 3. Download Gemma 4 26B. 4. Use Docker to keep everything contained. 5. Add Firecrawl so the agent can research the web. 6. Connect Telegram so you can chat with it from your phone. 7. Turn on context compression so it stays sharp. 8. Set the daily reset for 4 AM so it refreshes while you sleep. What you get at the end: → local AI agent → private workflow → web research built in → self improving skills → no monthly model rental That is a serious setup for anyone building with AI. Save this video, you’ll have the easiest beginner roadmap to follow later. Want the SOP? DM me. 💬

Julian Goldie SEO

31,292 просмотров • 3 месяцев назад

training a model that takes a text prompt and generates audio that renders video on an oscilloscope AgenC agents live inside worlds the model generates the pipeline: real videos -> edge detection -> vectorization -> path ordering -> 192kHz 3-channel WAV where X/Y control beam position and Z controls beam intensity 3 values per timestep. that's all the model is learning. compare that to video gen models trying to predict millions of pixels per frame. transformers are already great at sequence prediction and that's literally all this is. waveform generation the output IS the playback. generate the audio, feed it to a scope, it draws the scene in real-time. there's no rendering step. it's analog so there's no pixel grid. you get continuous curves and effectively infinite resolution bootstrapped with procedural data, lissajous curves, wireframe 3D, stick figures, then scaled on real-world video converted to trace format. 90 TB of source video the model learns edges, contours, spatial relationships, motion. once it has that, describing a scene it's never seen is novel trajectory through the same learned space. generative geometry

training a model that takes a text prompt and generates audio that renders video on an oscilloscope AgenC agents live inside worlds the model generates the pipeline: real videos -> edge detection -> vectorization -> path ordering -> 192kHz 3-channel WAV where X/Y control beam position and Z controls beam intensity 3 values per timestep. that's all the model is learning. compare that to video gen models trying to predict millions of pixels per frame. transformers are already great at sequence prediction and that's literally all this is. waveform generation the output IS the playback. generate the audio, feed it to a scope, it draws the scene in real-time. there's no rendering step. it's analog so there's no pixel grid. you get continuous curves and effectively infinite resolution bootstrapped with procedural data, lissajous curves, wireframe 3D, stick figures, then scaled on real-world video converted to trace format. 90 TB of source video the model learns edges, contours, spatial relationships, motion. once it has that, describing a scene it's never seen is novel trajectory through the same learned space. generative geometry

tetsuo

18,397 просмотров • 5 месяцев назад

We are excited to share our #ICRA2026 paper "Dream to Fly: Model-Based Reinforcement Learning for Vision-Based Drone Flight"! Paper: Video: Can we use Model-Based #ReinforcementLearning (MBRL) to fly a drone from pixels to commands? In this work, we train quadrotor navigation policies from scratch using #WorldModels, mapping raw onboard camera pixels directly to control commands, much like a human pilot! While model-free methods like PPO are sample-inefficient and struggle in this setting, we leverage #MBRL to train visuomotor policies capable of agile flight through a racetrack using only raw pixel observations, no explicit state estimation needed. A key finding: because our policies are trained end-to-end directly from pixels, we no longer need the perception-aware reward term used in previous methods. Instead, this behavior emerges naturally! The policies learn to guide the camera toward feature-rich areas of the observation space on their own. Kudos to Ángel Romero Ashwin Shenai Ismail Geles Elie Aljalbout Reference: "Dream to Fly: Model-Based Reinforcement Learning for Vision-Based Drone Flight" Angel Romero*, Ashwin Shenai*, Ismail Geles, Elie Aljalbout, Davide Scaramuzza IEEE International Conference on Robotics and Automation (ICRA), Vienna, 2026. European Research Council (ERC) AUTOASSESS UZH IfI University of Zurich UZH Science Prophesee SynSense UZH Space Hub Swiss Robotics NCCR Robotics

We are excited to share our #ICRA2026 paper "Dream to Fly: Model-Based Reinforcement Learning for Vision-Based Drone Flight"! Paper: Video: Can we use Model-Based #ReinforcementLearning (MBRL) to fly a drone from pixels to commands? In this work, we train quadrotor navigation policies from scratch using #WorldModels, mapping raw onboard camera pixels directly to control commands, much like a human pilot! While model-free methods like PPO are sample-inefficient and struggle in this setting, we leverage #MBRL to train visuomotor policies capable of agile flight through a racetrack using only raw pixel observations, no explicit state estimation needed. A key finding: because our policies are trained end-to-end directly from pixels, we no longer need the perception-aware reward term used in previous methods. Instead, this behavior emerges naturally! The policies learn to guide the camera toward feature-rich areas of the observation space on their own. Kudos to Ángel Romero Ashwin Shenai Ismail Geles Elie Aljalbout Reference: "Dream to Fly: Model-Based Reinforcement Learning for Vision-Based Drone Flight" Angel Romero, Ashwin Shenai, Ismail Geles, Elie Aljalbout, Davide Scaramuzza IEEE International Conference on Robotics and Automation (ICRA), Vienna, 2026. European Research Council (ERC) AUTOASSESS UZH IfI University of Zurich UZH Science Prophesee SynSense UZH Space Hub Swiss Robotics NCCR Robotics

Davide Scaramuzza

15,986 просмотров • 3 месяцев назад

Building AI agents and tools is getting ridiculously easy! Here is a pretty impressive FREE and Open-Source AI agent builder that runs on your computer. It's a desktop app you can install on Mac or Windows. Link: • Free • 100% no-code: You can type what you want, and the tool will build everything for you. • You can use local models like DeepSeek or Qwen, or connect to cloud models like GPT-4o, Claude, or Gemini with your API key. • You can turn the tools you build into MCP Servers. I recorded a quick video to show you how easy it is to build something from scratch.

Building AI agents and tools is getting ridiculously easy! Here is a pretty impressive FREE and Open-Source AI agent builder that runs on your computer. It's a desktop app you can install on Mac or Windows. Link: • Free • 100% no-code: You can type what you want, and the tool will build everything for you. • You can use local models like DeepSeek or Qwen, or connect to cloud models like GPT-4o, Claude, or Gemini with your API key. • You can turn the tools you build into MCP Servers. I recorded a quick video to show you how easy it is to build something from scratch.

Santiago

56,815 просмотров • 1 год назад

🚨 BREAKING: Big news in the computer vision world! 🎥 Luxonis | Robotic Vision just dropped its new OAK 4 line, and it’s a big upgrade for edge computer vision. Instead of being “just a stereo camera,” OAK 4 is a fully standalone vision computer with 52 TOPS of on-device AI. Models run locally, depth is computed locally, and no external PC or cloud pipeline is required. This is why robotics teams love it: lower latency, lower cost, fewer failure points in the field. The hardware is built for the real-world. IP67, shock-resistant, wide-FOV RGB + stereo pair, IR projection, IMU, audio, and a patent-pending calibration system that keeps depth accurate even when conditions change. But the real move is the platform. With Luxonis Hub, you can deploy models, grab telemetry, push OTA updates, or collect data when performance drifts, all from a unified interface. It turns a single device into an end-to-end edge CV system. Most customers today in robotics are groups who just want something that works: AMRs, bin-picking systems, trailer-loading robots, and ag-tech. 🤖 And they all say the same thing, the appeal isn’t raw TOPS, it’s the all-in-one simplicity that lets them scale without building custom infrastructure. Feels like the direction edge vision has been waiting for: rugged hardware + high-throughput on-device compute + a real management layer. A next step toward “plug-and-deploy” perception for robots. 🔗 Find out more here: ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

🚨 BREAKING: Big news in the computer vision world! 🎥 Luxonis | Robotic Vision just dropped its new OAK 4 line, and it’s a big upgrade for edge computer vision. Instead of being “just a stereo camera,” OAK 4 is a fully standalone vision computer with 52 TOPS of on-device AI. Models run locally, depth is computed locally, and no external PC or cloud pipeline is required. This is why robotics teams love it: lower latency, lower cost, fewer failure points in the field. The hardware is built for the real-world. IP67, shock-resistant, wide-FOV RGB + stereo pair, IR projection, IMU, audio, and a patent-pending calibration system that keeps depth accurate even when conditions change. But the real move is the platform. With Luxonis Hub, you can deploy models, grab telemetry, push OTA updates, or collect data when performance drifts, all from a unified interface. It turns a single device into an end-to-end edge CV system. Most customers today in robotics are groups who just want something that works: AMRs, bin-picking systems, trailer-loading robots, and ag-tech. 🤖 And they all say the same thing, the appeal isn’t raw TOPS, it’s the all-in-one simplicity that lets them scale without building custom infrastructure. Feels like the direction edge vision has been waiting for: rugged hardware + high-throughput on-device compute + a real management layer. A next step toward “plug-and-deploy” perception for robots. 🔗 Find out more here: ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

Lukas Ziegler

41,955 просмотров • 7 месяцев назад

Tired of teleoperation? One human video → 1,000s of robot demos. (📍GitHub ) Scaling Robot Data Without Dynamics Simulation or Robot Hardware Real2Render2Real (R2R2R) is a new way to scale robot data without physics simulation or hardware. You take a phone scan + a single monocular human demo. It tracks the motion, renders photorealistic scenes, and generates diverse, robot-agnostic trajectories ready for training. > No teleop, no sim, no robot, just a phone and a video > Train VLA models and diffusion policies directly on the output > Supports multiple robot embodiments with kinematic consistency > 1000s of demos in 1/27 the time of real-world collection Thank you, Max Fu, for sharing!! Project: Paper: Code coming soon: It shows that with the right pipeline, you can scale robot learning data without touching a robot. One of the most interesting directions in scalable robotics today. —— Weekly robotics and AI insights. Subscribe free:

Tired of teleoperation? One human video → 1,000s of robot demos. (📍GitHub ) Scaling Robot Data Without Dynamics Simulation or Robot Hardware Real2Render2Real (R2R2R) is a new way to scale robot data without physics simulation or hardware. You take a phone scan + a single monocular human demo. It tracks the motion, renders photorealistic scenes, and generates diverse, robot-agnostic trajectories ready for training. > No teleop, no sim, no robot, just a phone and a video > Train VLA models and diffusion policies directly on the output > Supports multiple robot embodiments with kinematic consistency > 1000s of demos in 1/27 the time of real-world collection Thank you, Max Fu, for sharing!! Project: Paper: Code coming soon: It shows that with the right pipeline, you can scale robot learning data without touching a robot. One of the most interesting directions in scalable robotics today. —— Weekly robotics and AI insights. Subscribe free:

Ilir Aliu

42,864 просмотров • 6 месяцев назад

I tried jack's Buzz. It's like Slack + OpenClaw + Herdr + but with some really unique features that people are sleeping on. The video below shows how it works, and some of my thoughts on the process and platform, e.g.: - Create and interact with agents on top of any harness (claude code, codex, pi, etc.) - Choose which models agents use, including local ones - Agents can delegate work and work in parallel in git worktrees - Agents are first-class citizens and work like humans (creating channels, delegating, access to chat history) - You can share AI compute within a community - It's completely open-source and decentralized Things I like: - Delegating work in chat feels natural: tag an agent, it replies in a thread with status updates as it e.g. compiles, commits, and deploys. - Shared compute: relay owners can share local compute with members, so a community could pool funds for one beefy machine running a local model and everyone uses it. - It's built on Nostr, an open protocol already tied into Bitcoin Lightning so I can imagine communities tipping each other or paying for compute/agent tasks with instant zero-fee micropayments in the future. - It ties together things like OpenClaw, an agent manager, and Slack-style chat into one tool. Things I didn't like: - You can't see what the agent is doing in a terminal. The activity view exists, but if you're used to watching a session run, this UI feels a bit abstracted. A terminal view would be great. - It feels slower than running a session in Claude Code, though no evidence to back that up. For that reason I found myself doing one-off tasks in the terminal instead. Verdict: - I really like it so far and can genuinely imagine working with a team this way. - It doesn't feel ready for big, complex tasks yet. For shallower tasks, it's perfect. - The shared compute + Nostr/Lightning angle is what really separates it from every other agent manager for me, and I think that future is coming.

I tried jack's Buzz. It's like Slack + OpenClaw + Herdr + but with some really unique features that people are sleeping on. The video below shows how it works, and some of my thoughts on the process and platform, e.g.: - Create and interact with agents on top of any harness (claude code, codex, pi, etc.) - Choose which models agents use, including local ones - Agents can delegate work and work in parallel in git worktrees - Agents are first-class citizens and work like humans (creating channels, delegating, access to chat history) - You can share AI compute within a community - It's completely open-source and decentralized Things I like: - Delegating work in chat feels natural: tag an agent, it replies in a thread with status updates as it e.g. compiles, commits, and deploys. - Shared compute: relay owners can share local compute with members, so a community could pool funds for one beefy machine running a local model and everyone uses it. - It's built on Nostr, an open protocol already tied into Bitcoin Lightning so I can imagine communities tipping each other or paying for compute/agent tasks with instant zero-fee micropayments in the future. - It ties together things like OpenClaw, an agent manager, and Slack-style chat into one tool. Things I didn't like: - You can't see what the agent is doing in a terminal. The activity view exists, but if you're used to watching a session run, this UI feels a bit abstracted. A terminal view would be great. - It feels slower than running a session in Claude Code, though no evidence to back that up. For that reason I found myself doing one-off tasks in the terminal instead. Verdict: - I really like it so far and can genuinely imagine working with a team this way. - It doesn't feel ready for big, complex tasks yet. For shallower tasks, it's perfect. - The shared compute + Nostr/Lightning angle is what really separates it from every other agent manager for me, and I think that future is coming.

Vinny

1,312,116 просмотров • 9 дней назад

In just one week, Binh Pham and I trained a full-body Unitree G1. Here's a recap: 1. Secured a Unitree G1 humanoid through a LinkedIn post 2. Deployed TWIST2 full-body teleoperation pipelines 3. Adapted TWIST2 for Zed stereo camera & collected full-body teleoperation samples (carried by Binh Pham ) 4. Adapted & fine-tuned NVIDIA Gr00T N1.5 VLA on the TWIST2 public datasets, which I fine-tuned on an 8xNVIDIA H100 Cluster. We picked Gr00T N1.5 as it was trained with Unitree G1 embodiment data. 5. Adapted the TWIST2 codebase to stream in the actions from Gr00T via ZMQ using a co-located NVIDIA H100 for ~200ms inference latency 6. Tested the model in sim, then deployed to the real-world Unitree G1. We streamed a training sample observation to the VLA (as we didn't want to break robot in case real observations were OOD) We were the first team in the world to deploy the full TWIST2 data collection pipeline to the unitree g1 :) Much more work ahead though, which I'll work on as a side-project over the next months: 1. Exploring the various types of 'world models': video backbones, dynamics models, v-jepa-2 models. I believe these will generalize better & train much more data-efficiently than VLM backbones 2. Speeding up inference - I believe low-latency robotics inference will be a big challenge. There are many works in video diffusion which I'd like to test (e.g. SageAttention, SparseAttention, Drifting Models). Perhaps also writing custom CUDA kernels. 3. Economics of inference scaling :) What will be the compute demands as we scale inference up to millions of humanoids? Will it run on edge or on distributed 'co-located' inference clusters? These are questions I'd like to answer. Adapted TWIST2 codebase: Adapted Gr00T-N1.5 codebase: The ETH Robotics Club are doing a cool GTC Golden ticket competition with NVIDIA , so this is my submission :) The DGX Spark compute will get me a long way with initial prototyping & especially working on inference optimization for next-gen Blackwell GPUs #NVIDIAGTC #GOLDENTICKET #ETHRC

In just one week, Binh Pham and I trained a full-body Unitree G1. Here's a recap: 1. Secured a Unitree G1 humanoid through a LinkedIn post 2. Deployed TWIST2 full-body teleoperation pipelines 3. Adapted TWIST2 for Zed stereo camera & collected full-body teleoperation samples (carried by Binh Pham ) 4. Adapted & fine-tuned NVIDIA Gr00T N1.5 VLA on the TWIST2 public datasets, which I fine-tuned on an 8xNVIDIA H100 Cluster. We picked Gr00T N1.5 as it was trained with Unitree G1 embodiment data. 5. Adapted the TWIST2 codebase to stream in the actions from Gr00T via ZMQ using a co-located NVIDIA H100 for ~200ms inference latency 6. Tested the model in sim, then deployed to the real-world Unitree G1. We streamed a training sample observation to the VLA (as we didn't want to break robot in case real observations were OOD) We were the first team in the world to deploy the full TWIST2 data collection pipeline to the unitree g1 :) Much more work ahead though, which I'll work on as a side-project over the next months: 1. Exploring the various types of 'world models': video backbones, dynamics models, v-jepa-2 models. I believe these will generalize better & train much more data-efficiently than VLM backbones 2. Speeding up inference - I believe low-latency robotics inference will be a big challenge. There are many works in video diffusion which I'd like to test (e.g. SageAttention, SparseAttention, Drifting Models). Perhaps also writing custom CUDA kernels. 3. Economics of inference scaling :) What will be the compute demands as we scale inference up to millions of humanoids? Will it run on edge or on distributed 'co-located' inference clusters? These are questions I'd like to answer. Adapted TWIST2 codebase: Adapted Gr00T-N1.5 codebase: The ETH Robotics Club are doing a cool GTC Golden ticket competition with NVIDIA , so this is my submission :) The DGX Spark compute will get me a long way with initial prototyping & especially working on inference optimization for next-gen Blackwell GPUs #NVIDIAGTC #GOLDENTICKET #ETHRC

Arnie Ramesh

14,815 просмотров • 5 месяцев назад

Agents: Quick thoughts & questions on how they operate, their potential, and their limitations A Few Observations - ▶️"Book me a hotel" or "pull historical financials" are already (mostly) solved problems!! Agents can do a ton of tasks right now—like parsing public company press releases and navigating capture key info & complete bookings accurately. However, for more complex navigation flows, the tech still needs some work - but I'm very confident it’s essentially a solved or solvable challenge. ▶️Accuracy & Speed - The key metrics and agents should optimize for. ▶️Lower Build & Migration Costs It took me two minutes to build a new website (link: This is great for consumers—more choices, lower switching costs. Companies will increasingly compete on the quality of their products and services. ▶️Agents vs. Automation tools: The more I think about it, the more I realize that most “agents” are really just automation tools—kind of like how most robots🤖 are just machines lol ❓A Few Key Questions—Would Love Your Thoughts! ❔Remote Servers & Logins In many cases, we’ll want agents to act on our behalf (e.g., log in to to cancel an order). How will platforms like respond? Many websites may block remote servers for security. Is there a technical workaround? ❔Agent Generalization Do we need to train agents on each environment separately, or can one solution handle multiple sites and systems? This seems similar to RL/post-training challenges in AI research. Example: It's unclear to me whether $Devin was specifically trained on environment? ❔Frontend vs. Backend infra for agents to run on I had doubts about Anthropic's "Computer Use" feature, which seemed to run on the frontend, basically remotely controlling my computer so I couldn’t use it at the same time. This should deliver the highest accuracy, but it’s questionable how practical it really is. (Ref: It seems def possible for agents to work quietly “in the background” (like Devin) rather than remotely controlling a user’s PC, but how much accuracy are we sacrificing? A few $Devin test cases that got me thinking: 1⃣Pulling $META's MAU and DAU (1Q21–3Q24) into Excel (video attached) Took Devin 11min - it sent me back an Excel with 100% accurate data. This case was pretty tricky because $Meta changed disclosures and stopped reporting MAU/DAU after 4Q23. Devin didn’t hallucinate data for post-4Q23—it simply didn’t provide it! It really shocked me to see $Devin navigating $Meta's investor relations site (I didn't tell it to find the numbers there), opening each quarterly earnings report, and extracting MAU/DAU like a diligent intern. -> This confirms $Devin (and similar agents) can already accurately “read” screens. 2⃣Booking Hotel (video attached) Devin took 5 minutes to book the InterContinental NYC on after asking for my credentials. From $Devin's workspace, I could see it filling in the correct fields and making the right selections—fast and accurate overall. Interestingly, $Devin didn’t supply all the required information on the first attempt and got some error messages, then retried until it succeeded. It’s unclear whether Devin had been specifically trained on interface or simply learned to adapt on the fly. 3⃣Canceling the Booking This part was even more interesting. While booking didn’t require me to log in, canceling did—so $Devin had to access my (likely via a remote server) account using my Gmail credentials. It successfully canceled the reservation. I wonder how websites will handle future “remote” logins. Notably, Google blocked $Devin’s direct attempts to log in to Gmail when I specifically requested it. 4⃣Booking from Official Hotel Sites I asked Devin to book InterContinental NYC and Four Seasons Boston via their official websites. It made progress but encountered technical hiccups when trying to select the check-in/check-out dates. Insights from Scott Wu on Invest Like the Best: 1/ Self-Driving Cars as the First “Real Agents” Driving requires near-perfect accuracy (99.999%), making it much more demanding than digital or coding agents, which can tolerate more errors. Scott compares $Devin to circa 2014—already good enough to save 90% of your effort, but still short of flawless. 2/ Impact on Collaboration Platforms Tools like Slack and GitLab are likely to see major changes as agents begin to interact with and utilize them along with humans. 2025 should be all about agents - both the disruptors and those they disrupt!

Agents: Quick thoughts & questions on how they operate, their potential, and their limitations A Few Observations - ▶️"Book me a hotel" or "pull historical financials" are already (mostly) solved problems!! Agents can do a ton of tasks right now—like parsing public company press releases and navigating capture key info & complete bookings accurately. However, for more complex navigation flows, the tech still needs some work - but I'm very confident it’s essentially a solved or solvable challenge. ▶️Accuracy & Speed - The key metrics and agents should optimize for. ▶️Lower Build & Migration Costs It took me two minutes to build a new website (link: This is great for consumers—more choices, lower switching costs. Companies will increasingly compete on the quality of their products and services. ▶️Agents vs. Automation tools: The more I think about it, the more I realize that most “agents” are really just automation tools—kind of like how most robots🤖 are just machines lol ❓A Few Key Questions—Would Love Your Thoughts! ❔Remote Servers & Logins In many cases, we’ll want agents to act on our behalf (e.g., log in to to cancel an order). How will platforms like respond? Many websites may block remote servers for security. Is there a technical workaround? ❔Agent Generalization Do we need to train agents on each environment separately, or can one solution handle multiple sites and systems? This seems similar to RL/post-training challenges in AI research. Example: It's unclear to me whether $Devin was specifically trained on environment? ❔Frontend vs. Backend infra for agents to run on I had doubts about Anthropic's "Computer Use" feature, which seemed to run on the frontend, basically remotely controlling my computer so I couldn’t use it at the same time. This should deliver the highest accuracy, but it’s questionable how practical it really is. (Ref: It seems def possible for agents to work quietly “in the background” (like Devin) rather than remotely controlling a user’s PC, but how much accuracy are we sacrificing? A few $Devin test cases that got me thinking: 1⃣Pulling $META's MAU and DAU (1Q21–3Q24) into Excel (video attached) Took Devin 11min - it sent me back an Excel with 100% accurate data. This case was pretty tricky because $Meta changed disclosures and stopped reporting MAU/DAU after 4Q23. Devin didn’t hallucinate data for post-4Q23—it simply didn’t provide it! It really shocked me to see $Devin navigating $Meta's investor relations site (I didn't tell it to find the numbers there), opening each quarterly earnings report, and extracting MAU/DAU like a diligent intern. -> This confirms $Devin (and similar agents) can already accurately “read” screens. 2⃣Booking Hotel (video attached) Devin took 5 minutes to book the InterContinental NYC on after asking for my credentials. From $Devin's workspace, I could see it filling in the correct fields and making the right selections—fast and accurate overall. Interestingly, $Devin didn’t supply all the required information on the first attempt and got some error messages, then retried until it succeeded. It’s unclear whether Devin had been specifically trained on interface or simply learned to adapt on the fly. 3⃣Canceling the Booking This part was even more interesting. While booking didn’t require me to log in, canceling did—so $Devin had to access my (likely via a remote server) account using my Gmail credentials. It successfully canceled the reservation. I wonder how websites will handle future “remote” logins. Notably, Google blocked $Devin’s direct attempts to log in to Gmail when I specifically requested it. 4⃣Booking from Official Hotel Sites I asked Devin to book InterContinental NYC and Four Seasons Boston via their official websites. It made progress but encountered technical hiccups when trying to select the check-in/check-out dates. Insights from Scott Wu on Invest Like the Best: 1/ Self-Driving Cars as the First “Real Agents” Driving requires near-perfect accuracy (99.999%), making it much more demanding than digital or coding agents, which can tolerate more errors. Scott compares $Devin to circa 2014—already good enough to save 90% of your effort, but still short of flawless. 2/ Impact on Collaboration Platforms Tools like Slack and GitLab are likely to see major changes as agents begin to interact with and utilize them along with humans. 2025 should be all about agents - both the disruptors and those they disrupt!

Freda Duan

48,850 просмотров • 1 год назад

this guy built an ai-girl pipeline using real-time face filters, and d2c brands now pay him $2,000 per ugc video he got tired of watching brands burn $4,000 on a single creator who takes 2 weeks to deliver one angle, so he built a setup that runs photoreal ai girls live from his own webcam, no actresses, no studios, no makeup artists his monthly revenue hit $89,000 last month from a network of 7 ai personas across tiktok and instagram. the average ugc creator caps at $6k juggling 4 brand deals the breakdown: > hardware is the moat, but most people butcher the setup in the first frame. face mesh locked at 60fps with zero artifacting > persona comes first, mess this up and nothing saves it: name, backstory, voice tone, niche before a single clip is shot > face selection is not random. you a/b test features (eye spacing, jawline, hair contrast) because some faces convert better in 9:16 > you're picking who your audience trusts, not who looks cool. that's targeting baked into bone structure > real-time physics run before the script, this is what kills the uncanny valley that destroys watch time in 2 seconds > the filter has to survive the strap of a tank top, the texture of a knit cardigan, the hair flick > batching is the move 96% skip: one performance, multiple personas, three platforms > the system pushes 12 pieces before lunch while brands test 2 creators a week and wonder why their cpa sits at $94 the economics: each video costs $4 in compute, sells for $1,500 to $3,000, takes 14 minutes to produce. that's a 37,500% margin, while ugc agencies pay creators $400-800 per clip and net $200 after revisions one supplement brand generated 14 variants with 7 personas in 4 hours and found a winner in 36 hours without flying a creator to la. they were paying $1,200 per ugc video and burning $6,000/week on content that didn't scale. now they spend $210 for 14 variants and their cpa dropped from $89 to $27 the avatars hold real products, warm window light on the persona, cold neon on the operator, mouth shapes sync to consonants not just vowels just a webcam, a tracked face, and the discipline to move enough that the filter never has a chance to break

Kiyoro

31,137 просмотров • 2 месяцев назад

X-Humanoid just officially dropped Embodied Tien Kung 3.0, A universal platform designed to be way more open and developer-friendly. 🤖 Built on their Wise Kaiwu AI platform, this next-gen humanoid is all about slashing development costs. It’s a fully interoperable ecosystem that supports everything from tactile interaction to high-dynamic motion control at a full humanoid scale. ➤ Radical Openness: X-Humanoid is open-sourcing the full stack—robot body, motion control, VLM/VLA models, and the RoboMIND dataset. It fully supports ROS2, MQTT, and TCP/IP, so developers can customize use cases without re-engineering the basics. ➤ High-Performance Hardware: With high-torque integrated joints, Tien Kung 3.0 can clear 1-meter (3.3ft) obstacles and handle dexterous moves like kneeling and bending. It hits millimeter-level precision, making it a solid fit for industrial-grade tasks. ➤ True Autonomy: The bot runs a continuous perception-decision-execution loop. It uses world models to break down complex language commands and VLA models for real-time obstacle avoidance and navigation. ➤ Scalable Collaboration: The platform moves beyond single-unit tasks to support multi-robot collaboration with autonomous scheduling. It’s built to move embodied AI from the lab straight into real-world commercial and industrial environments. Source: X-Humanoid #Humanoid #OpenSource #Robotics #EmbodiedAI #PhysicalAI #Automation #XHumanoid #TienKung #WiseKaiwu

X-Humanoid just officially dropped Embodied Tien Kung 3.0, A universal platform designed to be way more open and developer-friendly. 🤖 Built on their Wise Kaiwu AI platform, this next-gen humanoid is all about slashing development costs. It’s a fully interoperable ecosystem that supports everything from tactile interaction to high-dynamic motion control at a full humanoid scale. ➤ Radical Openness: X-Humanoid is open-sourcing the full stack—robot body, motion control, VLM/VLA models, and the RoboMIND dataset. It fully supports ROS2, MQTT, and TCP/IP, so developers can customize use cases without re-engineering the basics. ➤ High-Performance Hardware: With high-torque integrated joints, Tien Kung 3.0 can clear 1-meter (3.3ft) obstacles and handle dexterous moves like kneeling and bending. It hits millimeter-level precision, making it a solid fit for industrial-grade tasks. ➤ True Autonomy: The bot runs a continuous perception-decision-execution loop. It uses world models to break down complex language commands and VLA models for real-time obstacle avoidance and navigation. ➤ Scalable Collaboration: The platform moves beyond single-unit tasks to support multi-robot collaboration with autonomous scheduling. It’s built to move embodied AI from the lab straight into real-world commercial and industrial environments. Source: X-Humanoid #Humanoid #OpenSource #Robotics #EmbodiedAI #PhysicalAI #Automation #XHumanoid #TienKung #WiseKaiwu

RoboHub🤖

49,440 просмотров • 5 месяцев назад

Autonomous driving and obstacles avoidance at drift speeds, challenging the limits of what is possible! In this demo, our vehicle can be seen performing #autonomousdriving at very high speeds, causing it to both skid and drift at turns, while also avoiding obstacles. At such speeds, given the inherent dynamics of the vehicle platform used, it is very easy for the vehicle to topple. The #reinforcementlearning based motion planning and decision making framework that is being demoed here is tasked with ensuring obstacles avoidance without compromising on the speed, to an extent possible, and to drive the vehicle as fast as possible. This is evident towards the end of the video, where it can be seen that our vehicle avoided static obstacles while drifting.This demonstrates the level of sophistication and agility in our framework to ensure proper control of the #autonomousvehicles at high-speeds. The use cases are many; to begin with, our generic off-roads autonomous driving research focuses on enabling autonomous navigation in previously unknown and unseen environments, while ensuring mathematical completeness guarantees. Such agility can also help on-road autonomous vehicles to deal with unforeseeable corner cases or sudden appearance of obstacles in its tracks, at high-speeds. Our underlying research at Swaayatt Robots is still far from over, and over the next 3-4 months, we will be demonstrating abstract representation being learned by our multi-RL agents based framework (under progress) to ensure computation of the cost of the terrain without any labelled data, where multiple agents learn to control / regulate different aspects of autonomous navigation, to ensure safe and robust navigation, both on- and off-roads. All the people on the ground, who participated in the demo, were trained safety professionals. #deeplearning #MachineLearning #Robotics

Autonomous driving and obstacles avoidance at drift speeds, challenging the limits of what is possible! In this demo, our vehicle can be seen performing #autonomousdriving at very high speeds, causing it to both skid and drift at turns, while also avoiding obstacles. At such speeds, given the inherent dynamics of the vehicle platform used, it is very easy for the vehicle to topple. The #reinforcementlearning based motion planning and decision making framework that is being demoed here is tasked with ensuring obstacles avoidance without compromising on the speed, to an extent possible, and to drive the vehicle as fast as possible. This is evident towards the end of the video, where it can be seen that our vehicle avoided static obstacles while drifting.This demonstrates the level of sophistication and agility in our framework to ensure proper control of the #autonomousvehicles at high-speeds. The use cases are many; to begin with, our generic off-roads autonomous driving research focuses on enabling autonomous navigation in previously unknown and unseen environments, while ensuring mathematical completeness guarantees. Such agility can also help on-road autonomous vehicles to deal with unforeseeable corner cases or sudden appearance of obstacles in its tracks, at high-speeds. Our underlying research at Swaayatt Robots is still far from over, and over the next 3-4 months, we will be demonstrating abstract representation being learned by our multi-RL agents based framework (under progress) to ensure computation of the cost of the terrain without any labelled data, where multiple agents learn to control / regulate different aspects of autonomous navigation, to ensure safe and robust navigation, both on- and off-roads. All the people on the ground, who participated in the demo, were trained safety professionals. #deeplearning #MachineLearning #Robotics

Sanjeev Sharma

11,181 просмотров • 1 год назад

This guy deserves to get Famous .. Such an amazing achievement at such young age .. The AirPods Feature Apple Quietly Took Away From Android Users Most companies build ecosystems. Apple builds walls. If you use AirPods with an Android phone, you already know this. Noise cancellation disabled. Ear detection gone. Features you paid for, locked behind a device you don't own. Not because the hardware can't do it because Apple decided it shouldn't. For most people, that's just the cost of choosing Android. Something you accept and move on from. Kavish Dewar didn't move on. In this video, we cover the full story: 1. How Apple restricts AirPods features on Android and why it is a deliberate design choice, not a technical limitation 2. How a 16-year-old high school student from Gurugram reverse-engineered that restriction system with no team, no funding, and no formal training 3. What LibrePods actually does and why it works so effectively 4. How the project went from a bedroom in Gurugram to GitHub's trending page to global coverage by The Verge 5. Why Kavish made LibrePods completely free and open source and what that decision says about the difference between building for impact versus building for profit Why this story matters beyond the tech: LibrePods is not just a clever app. It is a direct challenge to the idea that trillion dollar companies get to define the limits of what you can do with hardware you already own. One teenager decided that was worth fighting not with lawyers, not with a startup, but with clean code and an open license. That is a different kind of power. And it is available to anyone willing to use it.

This guy deserves to get Famous .. Such an amazing achievement at such young age .. The AirPods Feature Apple Quietly Took Away From Android Users Most companies build ecosystems. Apple builds walls. If you use AirPods with an Android phone, you already know this. Noise cancellation disabled. Ear detection gone. Features you paid for, locked behind a device you don't own. Not because the hardware can't do it because Apple decided it shouldn't. For most people, that's just the cost of choosing Android. Something you accept and move on from. Kavish Dewar didn't move on. In this video, we cover the full story: 1. How Apple restricts AirPods features on Android and why it is a deliberate design choice, not a technical limitation 2. How a 16-year-old high school student from Gurugram reverse-engineered that restriction system with no team, no funding, and no formal training 3. What LibrePods actually does and why it works so effectively 4. How the project went from a bedroom in Gurugram to GitHub's trending page to global coverage by The Verge 5. Why Kavish made LibrePods completely free and open source and what that decision says about the difference between building for impact versus building for profit Why this story matters beyond the tech: LibrePods is not just a clever app. It is a direct challenge to the idea that trillion dollar companies get to define the limits of what you can do with hardware you already own. One teenager decided that was worth fighting not with lawyers, not with a startup, but with clean code and an open license. That is a different kind of power. And it is available to anyone willing to use it.

AstroCounselKK 🇮🇳

31,158 просмотров • 4 месяцев назад

Matthew Gallagher Built a $401M Company in Year One with 2 People. And the tool behind it? Claude Code. This year he's on track for $1.8B. Sam Altman predicted this. It's happening now. The problem? It costs money. API credits stack up. Monthly bills keep growing. Every prompt eats your budget. Every project drains your wallet faster. Until now. Two methods. 99% cheaper. One is completely free. Forever. $0. Not a trial. This video breaks down both step by step. ↓ Let me put this in perspective. $100-$500. That's monthly. That's what you spend. That's $6,000/year on API credits. Just to use a tool you haven't shipped anything with. The $401M guy? Spending $0. Same capability. Shipping weekly. Different cost structure. Different results. Different life. I'm about to hand you his cost structure for free. ↓ Open source vs closed source. Pay attention. Closed source: Claude. GPT-4. Pay per token. Meter always running. Open source: Qwen. Llama. Mistral. Free to download. Free to run. Free forever. No meter. No tokens. No bill. Here's what nobody tells you: 80% of coding tasks? Open source handles them. More than handles them. Writes clean code. Debugs errors. Generates boilerplate. Handles routine work perfectly. You're paying premium prices for tasks that don't need premium intelligence. That's hiring a brain surgeon to put on a bandaid. Smart play: Free models for the 80%. Paid credits for the 20%. That's what the $401M guy does. That's what this video teaches you. Follow Himanshu Kumar for more breakdowns that turn free tools into real businesses. ↓ Method 1: Ollama. Local. Free. Forever. Download it. Pull a model. Point Claude Code at it. Done. No internet needed. No API keys required. No monthly subscription. No token counting ever. No bill. Today. Tomorrow. Ever. Your data never leaves your computer. Complete privacy. Complete freedom. Claude Code thinks it's talking to the cloud. It's talking to your laptop. For $0. The video walks through every step: Every config file. Every variable. Every command. Every click. If you can follow a recipe, you can do this. People who set this up 3 months ago? Saved $300-$1,500 since then. Workflow didn't change one bit. ↓ Hardware you need: 16GB RAM: 7B models run smooth. 32GB RAM: 32B models run comfortable. 64GB + GPU: biggest models available. No GPU? Still works. Just slower. Few extra seconds. That's it. Your $1,500 laptop is sitting there running Chrome and Spotify. Put it to work saving you $200/month instead. Follow Himanshu Kumar for more breakdowns that turn free tools into real businesses. ↓ Method 2: Open Router. Free Cloud. No Hardware. Weak machine? Don't want local setup? This method is for you. Free AI models in the cloud. No download. No hardware. Configure Claude Code to route through Open Router. The config: Base URL: Open Router API. API key: free Open Router key. Default Sonnet: free. Default Opus: free. Default Haiku: free. Small fast model: free. Subagent model: free. Free. Free. Free. Free. Free across the board. Same interface. Same commands. Same workflow. Zero cost. Copy the config from the video. Paste it. Save $200/month. Starting today. Right now. ↓ When to use which: Ollama (local): Best for privacy. Best for offline work. Best for unlimited usage. Best if you have decent hardware. Open Router (cloud): Best for weak machines. Best for instant setup. Best for trying different models. Best if you don't want to manage anything. Both methods: Best for 80% of your daily work. Still use paid Claude for: Complex architecture. Multi-file refactoring. Deep reasoning tasks. The 20% that actually needs it. $20/month instead of $200/month. Same output. 90% less cost. ↓ The math that should make you angry. You (current): $200-$500/month. $2,400-$6,000/year. $7,200-$18,000 over 3 years. You (after this video): $20-$50/month. $240-$600/year. $720-$1,800 over 3 years. Savings over 3 years: $6,480-$16,200. That's a used car. That's seed money. That's 6 months of rent. All from one 25-minute video. All from 15 minutes of configuration. Highest ROI 25 minutes you'll spend this year. ↓ The limitations. I won't lie to you. Open source is not Opus. Not as smart on complex reasoning. Not as good at long-context tasks. Makes more mistakes on nuanced problems. But they are: Free. Capable. Getting better monthly. Good enough for 80% of daily work. Smart cost management isn't being cheap. It's being strategic. Expensive tool when it matters. Free tool when it doesn't. ↓ The one-person billion-dollar company is coming. $401M in year one proved it's possible. The building blocks: AI that codes: Claude Code. Way to run it free: this video. Distribution: the internet. Customers: everyone. Only missing ingredient? Someone who builds. Not reads about building. Not saves posts about building. Not bookmarks videos about building. Builds. Tools are free. Knowledge is free. Opportunity is screaming. You're still "thinking about it." ↓ Your action plan: Tonight: Watch the video. Tomorrow morning: Set up Ollama or Open Router. Tomorrow afternoon: Build something. Anything. This week: Build a second thing. Faster. This month: Charge someone for it. One video. One setup. One weekend. $0 cost. Unlimited potential. Or keep paying $200/month for something you could get free. Keep consuming instead of building. Keep planning instead of shipping. Matthew Gallagher didn't plan a $401M company. He built it. Full video attached. Every method. Every config. Every tradeoff. 25 minutes. Your move. Follow Himanshu Kumar for more breakdowns that turn free tools into real businesses.

Matthew Gallagher Built a $401M Company in Year One with 2 People. And the tool behind it? Claude Code. This year he's on track for $1.8B. Sam Altman predicted this. It's happening now. The problem? It costs money. API credits stack up. Monthly bills keep growing. Every prompt eats your budget. Every project drains your wallet faster. Until now. Two methods. 99% cheaper. One is completely free. Forever. $0. Not a trial. This video breaks down both step by step. ↓ Let me put this in perspective. $100-$500. That's monthly. That's what you spend. That's $6,000/year on API credits. Just to use a tool you haven't shipped anything with. The $401M guy? Spending $0. Same capability. Shipping weekly. Different cost structure. Different results. Different life. I'm about to hand you his cost structure for free. ↓ Open source vs closed source. Pay attention. Closed source: Claude. GPT-4. Pay per token. Meter always running. Open source: Qwen. Llama. Mistral. Free to download. Free to run. Free forever. No meter. No tokens. No bill. Here's what nobody tells you: 80% of coding tasks? Open source handles them. More than handles them. Writes clean code. Debugs errors. Generates boilerplate. Handles routine work perfectly. You're paying premium prices for tasks that don't need premium intelligence. That's hiring a brain surgeon to put on a bandaid. Smart play: Free models for the 80%. Paid credits for the 20%. That's what the $401M guy does. That's what this video teaches you. Follow Himanshu Kumar for more breakdowns that turn free tools into real businesses. ↓ Method 1: Ollama. Local. Free. Forever. Download it. Pull a model. Point Claude Code at it. Done. No internet needed. No API keys required. No monthly subscription. No token counting ever. No bill. Today. Tomorrow. Ever. Your data never leaves your computer. Complete privacy. Complete freedom. Claude Code thinks it's talking to the cloud. It's talking to your laptop. For $0. The video walks through every step: Every config file. Every variable. Every command. Every click. If you can follow a recipe, you can do this. People who set this up 3 months ago? Saved $300-$1,500 since then. Workflow didn't change one bit. ↓ Hardware you need: 16GB RAM: 7B models run smooth. 32GB RAM: 32B models run comfortable. 64GB + GPU: biggest models available. No GPU? Still works. Just slower. Few extra seconds. That's it. Your $1,500 laptop is sitting there running Chrome and Spotify. Put it to work saving you $200/month instead. Follow Himanshu Kumar for more breakdowns that turn free tools into real businesses. ↓ Method 2: Open Router. Free Cloud. No Hardware. Weak machine? Don't want local setup? This method is for you. Free AI models in the cloud. No download. No hardware. Configure Claude Code to route through Open Router. The config: Base URL: Open Router API. API key: free Open Router key. Default Sonnet: free. Default Opus: free. Default Haiku: free. Small fast model: free. Subagent model: free. Free. Free. Free. Free. Free across the board. Same interface. Same commands. Same workflow. Zero cost. Copy the config from the video. Paste it. Save $200/month. Starting today. Right now. ↓ When to use which: Ollama (local): Best for privacy. Best for offline work. Best for unlimited usage. Best if you have decent hardware. Open Router (cloud): Best for weak machines. Best for instant setup. Best for trying different models. Best if you don't want to manage anything. Both methods: Best for 80% of your daily work. Still use paid Claude for: Complex architecture. Multi-file refactoring. Deep reasoning tasks. The 20% that actually needs it. $20/month instead of $200/month. Same output. 90% less cost. ↓ The math that should make you angry. You (current): $200-$500/month. $2,400-$6,000/year. $7,200-$18,000 over 3 years. You (after this video): $20-$50/month. $240-$600/year. $720-$1,800 over 3 years. Savings over 3 years: $6,480-$16,200. That's a used car. That's seed money. That's 6 months of rent. All from one 25-minute video. All from 15 minutes of configuration. Highest ROI 25 minutes you'll spend this year. ↓ The limitations. I won't lie to you. Open source is not Opus. Not as smart on complex reasoning. Not as good at long-context tasks. Makes more mistakes on nuanced problems. But they are: Free. Capable. Getting better monthly. Good enough for 80% of daily work. Smart cost management isn't being cheap. It's being strategic. Expensive tool when it matters. Free tool when it doesn't. ↓ The one-person billion-dollar company is coming. $401M in year one proved it's possible. The building blocks: AI that codes: Claude Code. Way to run it free: this video. Distribution: the internet. Customers: everyone. Only missing ingredient? Someone who builds. Not reads about building. Not saves posts about building. Not bookmarks videos about building. Builds. Tools are free. Knowledge is free. Opportunity is screaming. You're still "thinking about it." ↓ Your action plan: Tonight: Watch the video. Tomorrow morning: Set up Ollama or Open Router. Tomorrow afternoon: Build something. Anything. This week: Build a second thing. Faster. This month: Charge someone for it. One video. One setup. One weekend. $0 cost. Unlimited potential. Or keep paying $200/month for something you could get free. Keep consuming instead of building. Keep planning instead of shipping. Matthew Gallagher didn't plan a $401M company. He built it. Full video attached. Every method. Every config. Every tradeoff. 25 minutes. Your move. Follow Himanshu Kumar for more breakdowns that turn free tools into real businesses.

Himanshu Kumar

13,599 просмотров • 3 месяцев назад

Claude + Obsidian + n8n built AI models that now clear $5,400 a month. Most builders chase the next big model and burn cash on cloud bills. This stack runs everything local and compounds your own work. The base is Karpathy’s 1-file method: one giant markdown note in Obsidian. New thoughts, datasets, training logs, and prompt tests dump at the top. Old ones sink. Every few days you reread and pull the survivors back up. No folders. No tags. No plugins. The rereading is the system. The flaw hits past 10,000 lines. No human rereads that. Claude takes over. Point it at the vault folder. Ask: “What worked on my last fine-tune.” “Find the 3 datasets I keep circling.” “What hyperparameters failed in February.” It answers from your own notes with exact quotes in 15 seconds. n8n closes the loop on autopilot. Once a week it triggers Claude: read the last 7 days, surface the 5 experiments worth pulling up, flag contradictions, then spin up the next training job. The model stays alive. You never chase old logs again. Week 1 feels like nothing. Week 4 you hit the first “I already solved this drift in January.” Month 3 you ship custom agents and fine-tunes faster than most teams. Most builders lose ideas in scattered notebooks. This one compounds because dumping takes zero discipline. You use the same system to create high-quality AI models, package them as agents, and sell them on marketplaces or to clients. $5,000+ a month becomes the floor once the vault hits critical mass. Notion stores what you thought. This thing ships models that argue back and pay rent.

Claude + Obsidian + n8n built AI models that now clear $5,400 a month. Most builders chase the next big model and burn cash on cloud bills. This stack runs everything local and compounds your own work. The base is Karpathy’s 1-file method: one giant markdown note in Obsidian. New thoughts, datasets, training logs, and prompt tests dump at the top. Old ones sink. Every few days you reread and pull the survivors back up. No folders. No tags. No plugins. The rereading is the system. The flaw hits past 10,000 lines. No human rereads that. Claude takes over. Point it at the vault folder. Ask: “What worked on my last fine-tune.” “Find the 3 datasets I keep circling.” “What hyperparameters failed in February.” It answers from your own notes with exact quotes in 15 seconds. n8n closes the loop on autopilot. Once a week it triggers Claude: read the last 7 days, surface the 5 experiments worth pulling up, flag contradictions, then spin up the next training job. The model stays alive. You never chase old logs again. Week 1 feels like nothing. Week 4 you hit the first “I already solved this drift in January.” Month 3 you ship custom agents and fine-tunes faster than most teams. Most builders lose ideas in scattered notebooks. This one compounds because dumping takes zero discipline. You use the same system to create high-quality AI models, package them as agents, and sell them on marketplaces or to clients. $5,000+ a month becomes the floor once the vault hits critical mass. Notion stores what you thought. This thing ships models that argue back and pay rent.

Rugikk

12,599 просмотров • 15 дней назад

We are thrilled to share our breakthrough research on "Agile Flight from Pixels without State Estimation," to be presented and live-demonstrated at #RSS2024 next week! You heard well: no state estimation means no explicit visual localization, no SLAM, no VIO, and no IMU! Paper: Video (Narrated): Last year, we demonstrated that #ReinforcementLearning (RL) policies could outperform world-champion drone-racing pilots using the same quadrotor hardware; however, unlike human pilots, these policies continuously estimated an explicit state from known gate positions, the camera feed, and inertial measurements (IMU). In this new work, we tackle the challenge of learning vision-based drone racing using an end-to-end reinforcement learning approach that eliminates the need for IMU data or explicit state estimation. Like professional pilots, we go directly from images to control commands. The training is facilitated by an asymmetric actor-critic with access to privileged information. To overcome the computational complexity during image-based RL training, we use an appropriate sensor representation, which can be efficiently simulated during training without rendering images. We achieve agile flight at speeds up to 40 km/h with accelerations up to 2 g's. Although our demonstration focuses on drone racing, we believe that our method has an impact beyond drone racing and can serve as a foundation for future research into real-world applications in structured environments. Besides the paper presentation, we will also give a live demo next Tuesday and Wednesday between and hrs at TU Delft: Reference: Ismail Geles*, Leonard Bauersfeld*, Angel Romero, Jiaxu Xing, Davide Scaramuzza "Demonstrating Agile Flight from Pixels without State Estimation" Robotics: Science and Systems (RSS), 2024. Kudos to Ismail Geles Leonard Bauersfeld Ángel Romero Jiaxu Xing! University of Zurich UZH Science UZH Space Hub Aerial Core AUTOASSESS European Research Council (ERC)

We are thrilled to share our breakthrough research on "Agile Flight from Pixels without State Estimation," to be presented and live-demonstrated at #RSS2024 next week! You heard well: no state estimation means no explicit visual localization, no SLAM, no VIO, and no IMU! Paper: Video (Narrated): Last year, we demonstrated that #ReinforcementLearning (RL) policies could outperform world-champion drone-racing pilots using the same quadrotor hardware; however, unlike human pilots, these policies continuously estimated an explicit state from known gate positions, the camera feed, and inertial measurements (IMU). In this new work, we tackle the challenge of learning vision-based drone racing using an end-to-end reinforcement learning approach that eliminates the need for IMU data or explicit state estimation. Like professional pilots, we go directly from images to control commands. The training is facilitated by an asymmetric actor-critic with access to privileged information. To overcome the computational complexity during image-based RL training, we use an appropriate sensor representation, which can be efficiently simulated during training without rendering images. We achieve agile flight at speeds up to 40 km/h with accelerations up to 2 g's. Although our demonstration focuses on drone racing, we believe that our method has an impact beyond drone racing and can serve as a foundation for future research into real-world applications in structured environments. Besides the paper presentation, we will also give a live demo next Tuesday and Wednesday between and hrs at TU Delft: Reference: Ismail Geles, Leonard Bauersfeld, Angel Romero, Jiaxu Xing, Davide Scaramuzza "Demonstrating Agile Flight from Pixels without State Estimation" Robotics: Science and Systems (RSS), 2024. Kudos to Ismail Geles Leonard Bauersfeld Ángel Romero Jiaxu Xing! University of Zurich UZH Science UZH Space Hub Aerial Core AUTOASSESS European Research Council (ERC)

Davide Scaramuzza

27,917 просмотров • 2 лет назад

What’s a “conservative crack up”? Its the lynchpin of an op being run on us today And it hails from an old playbook: “The Case for Paleo Libertarianism" It declares that a conservative “crack up” represents an historic opportunity for the libertarian movement. How? ➡️ Reject “cosmopolitan elites” (Epstein class) ➡️ Undermine trust in centralized institutions (DOJ, FBI, COA) ➡️ Align with traditional cultural values ➡️ “Burn it down” ➡️Build a new coalition outside the political mainstream (see below) They’ve got to convince us that Trump had betrayed and failed us - even tho it’s less than 1.5 years into a 4 year term- bc otherwise their plan fails. They’re coordinated and they’re misleading us. They’re going to attempt a run at 2028 but they need MAGA out of the way to succeed. As far as I’m concerned they’ve already lost. 📽️ Video 1 (2013) Ron Paul speaks on the paleo-libertarian playbook & lauded what he called the “grand coalition” of the libertarian Republicans and the Democratic progressives that have joined together in opposition to the use of force. 📽️Video 2 (2026) MTG discusses coordination w Ro Khanna & others to pull together a coalition from the left and the right to form a new center bc the current one has failed us 📽️Video 3 (2026) Joe Kent says he believes we need a new bipartisan movement that charts a new path. He says he’d like to be part of that administration in 2028 of the chance presents itself (LOL) 📱pic (2025) Thomas Massie forms an alliance w Ro Khanna re Epstein related matters

What’s a “conservative crack up”? Its the lynchpin of an op being run on us today And it hails from an old playbook: “The Case for Paleo Libertarianism" It declares that a conservative “crack up” represents an historic opportunity for the libertarian movement. How? ➡️ Reject “cosmopolitan elites” (Epstein class) ➡️ Undermine trust in centralized institutions (DOJ, FBI, COA) ➡️ Align with traditional cultural values ➡️ “Burn it down” ➡️Build a new coalition outside the political mainstream (see below) They’ve got to convince us that Trump had betrayed and failed us - even tho it’s less than 1.5 years into a 4 year term- bc otherwise their plan fails. They’re coordinated and they’re misleading us. They’re going to attempt a run at 2028 but they need MAGA out of the way to succeed. As far as I’m concerned they’ve already lost. 📽️ Video 1 (2013) Ron Paul speaks on the paleo-libertarian playbook & lauded what he called the “grand coalition” of the libertarian Republicans and the Democratic progressives that have joined together in opposition to the use of force. 📽️Video 2 (2026) MTG discusses coordination w Ro Khanna & others to pull together a coalition from the left and the right to form a new center bc the current one has failed us 📽️Video 3 (2026) Joe Kent says he believes we need a new bipartisan movement that charts a new path. He says he’d like to be part of that administration in 2028 of the chance presents itself (LOL) 📱pic (2025) Thomas Massie forms an alliance w Ro Khanna re Epstein related matters

SheThinksFreely

20,436 просмотров • 3 месяцев назад

SORA by Hand ✍️ OpenAI’s #SORA took over the Internet when it was announced earlier this year. The technology behind Sora is the Diffusion Transformer (DiT) developed by William Peebles and Shining Xie. How does DiT work? 𝗚𝗼𝗮𝗹: Generate a video conditioned by a text prompt and a series of diffusion steps [1] Given ↳ Video ↳ Prompt: "sora is sky" ↳ Diffusion step: t = 3 [2] Video → Patches ↳ Divide all pixels in all frames into 4 spacetime patches [3] Visual Encoder: Pixels 🟨 → Latent 🟩 ↳ Multiply the patches with weights and biases, followed by ReLU ↳ The result is a latent feature vector per patch ↳ The purpose is dimension reduction from 4 (2x2x1) to 2 (2x1). ↳ In the paper, the reduction is 196,608 (256x256x3)→ 4096 (32x32x4) [4] ⬛ Add Noise ↳ Sample a noise according to the diffusion time step t. Typically, the larger the t, the smaller the noise. ↳ Add the Sampled Noise to latent features to obtain Noised Latent. ↳ The goal is to purposely add noise to a video and ask the model to guess what that noise is. ↳ This is analogous to training a language model by purposely deleting a word in a sentence and ask the model to guess what the deleted word was. [5-7] 🟪 Conditioning by Adaptive Layer Norm [5] Encode Conditions ↳ Encode "sora is sky" into a text embedding vector [0,1,-1]. ↳ Encode t = 3 to as a binary vector [1,1]. ↳ Concatenate the two vectors in to a 5D column vector. [6] Estimate Scale/Shift ↳ Multiply the combined vector with weights and biases ↳ The goal is to estimate the scale [2,-1] and shift [-1,5]. ↳ Copy the result to (X) and (+) [7] Apply Scale/Sift ↳ Scale the noised latent by [2,-1] ↳ Shifted the scaled noised latent by [-1, 5] ↳ The result is "conditioned" noise latent. [8-10] Transformer [8] Self-Attention ↳ Feed the conditioned noised latent to Query-Key function to obtain a self-attention matrix ↳ Value is omitted for simplicity [9] Attention Pooling ↳ Multiply the conditioned noised latent with the self-attention matrix ↳ The result are attention weighted features [10] Pointwise Feed Forward Network ↳ Multiply the attention weighted features with weights and biases ↳ The result is the Predicted Noise 🏋️‍♂️ 𝗧𝗿𝗮𝗶𝗻 [11] ↳ Calculate MSE loss gradients by taking the different between the Predicted Noise and the Sampled Noise (ground truth). ↳ Use the loss gradients to kick off backpropagation to update all learnable parameters (red borders) ↳ Note the visual encoder and decoder's parameters are frozen (blue borders) 🎨 𝗚𝗲𝗻𝗲𝗿𝗮𝘁𝗲 (𝗦𝗮𝗺𝗽𝗹𝗲) [12] Denoise ↳ Subtract the predicted noise from the noised latent to obtain the noise-free latent [13] Visual Decoder: Latent 🟩 → Pixels 🟨 ↳ Multiply the patches with weights and biases, followed by ReLU [14] Patches → Video ↳ Rearrange patches into a sequence of video frames.

SORA by Hand ✍️ OpenAI’s #SORA took over the Internet when it was announced earlier this year. The technology behind Sora is the Diffusion Transformer (DiT) developed by William Peebles and Shining Xie. How does DiT work? 𝗚𝗼𝗮𝗹: Generate a video conditioned by a text prompt and a series of diffusion steps [1] Given ↳ Video ↳ Prompt: "sora is sky" ↳ Diffusion step: t = 3 [2] Video → Patches ↳ Divide all pixels in all frames into 4 spacetime patches [3] Visual Encoder: Pixels 🟨 → Latent 🟩 ↳ Multiply the patches with weights and biases, followed by ReLU ↳ The result is a latent feature vector per patch ↳ The purpose is dimension reduction from 4 (2x2x1) to 2 (2x1). ↳ In the paper, the reduction is 196,608 (256x256x3)→ 4096 (32x32x4) [4] ⬛ Add Noise ↳ Sample a noise according to the diffusion time step t. Typically, the larger the t, the smaller the noise. ↳ Add the Sampled Noise to latent features to obtain Noised Latent. ↳ The goal is to purposely add noise to a video and ask the model to guess what that noise is. ↳ This is analogous to training a language model by purposely deleting a word in a sentence and ask the model to guess what the deleted word was. [5-7] 🟪 Conditioning by Adaptive Layer Norm [5] Encode Conditions ↳ Encode "sora is sky" into a text embedding vector [0,1,-1]. ↳ Encode t = 3 to as a binary vector [1,1]. ↳ Concatenate the two vectors in to a 5D column vector. [6] Estimate Scale/Shift ↳ Multiply the combined vector with weights and biases ↳ The goal is to estimate the scale [2,-1] and shift [-1,5]. ↳ Copy the result to (X) and (+) [7] Apply Scale/Sift ↳ Scale the noised latent by [2,-1] ↳ Shifted the scaled noised latent by [-1, 5] ↳ The result is "conditioned" noise latent. [8-10] Transformer [8] Self-Attention ↳ Feed the conditioned noised latent to Query-Key function to obtain a self-attention matrix ↳ Value is omitted for simplicity [9] Attention Pooling ↳ Multiply the conditioned noised latent with the self-attention matrix ↳ The result are attention weighted features [10] Pointwise Feed Forward Network ↳ Multiply the attention weighted features with weights and biases ↳ The result is the Predicted Noise 🏋️‍♂️ 𝗧𝗿𝗮𝗶𝗻 [11] ↳ Calculate MSE loss gradients by taking the different between the Predicted Noise and the Sampled Noise (ground truth). ↳ Use the loss gradients to kick off backpropagation to update all learnable parameters (red borders) ↳ Note the visual encoder and decoder's parameters are frozen (blue borders) 🎨 𝗚𝗲𝗻𝗲𝗿𝗮𝘁𝗲 (𝗦𝗮𝗺𝗽𝗹𝗲) [12] Denoise ↳ Subtract the predicted noise from the noised latent to obtain the noise-free latent [13] Visual Decoder: Latent 🟩 → Pixels 🟨 ↳ Multiply the patches with weights and biases, followed by ReLU [14] Patches → Video ↳ Rearrange patches into a sequence of video frames.

Tom Yeh

238,241 просмотров • 2 лет назад

Most video-action robot models are a content-creation video generator with an action module attached. LingBot-VA 2.0 from Robbyant, a video-action foundation model, throws that starting point out and trains the whole stack natively for control. And it runs closed-loop at a peak 225 Hz. It's so important because A robot cannot move responsively when its controller pauses to imagine the next few frames. LingBot-VA 2.0 predicts during execution, then corrects using each real observation. And it carries only about 13B video parameters while activating roughly 1.9B per token. Bigger robot models usually mean slower reactions, creating a direct conflict between intelligence and control. LingBot-VA 2.0 is trained from scratch for robot control rather than adapted from a video generator built for content creation. Robbyant, an embodied AI company under Ant Group, built it to learn how scenes change under actions, predict what should happen next, and turn those predictions into real-time robot movements. Most video-action systems inherit a tokenizer and video backbone trained mainly to reproduce visual appearance. LingBot-VA 2.0 rebuilds both parts around physical control. Its semantic visual-action tokenizer maps observations toward features from a frozen vision foundation model and learns compact latent actions from frame-to-frame changes using self-supervised inverse and forward dynamics. Unlabeled web video can therefore carry action-relevant training signals without robot action labels. The policy is causal from the start, so every prediction can use only past observations. Its sparse Mixture-of-Experts video backbone has about 13B total parameters, while about 1.9B are active per token, keeping the compute lower during each step. A high-level vision-language planner breaks long tasks into smaller instructions, while the low-level video-action policy handles continuous movement. Foresight Reasoning predicts future visual states while the robot is already acting, then replaces imagined states with every new real observation. Combined with few-step distillation and systems acceleration, the paper reports a peak asynchronous execution frequency of 225 Hz. The model adapts from 10–15 demonstrations, transfers across robot embodiments, and handles some new tasks zero-shot. In the paper’s own evaluations, it reaches 93.6 average on RoboTwin 2.0 and reports stronger real-world results than LingBot-VA and π0.5 across the tested tasks. 🧵 1.

Most video-action robot models are a content-creation video generator with an action module attached. LingBot-VA 2.0 from Robbyant, a video-action foundation model, throws that starting point out and trains the whole stack natively for control. And it runs closed-loop at a peak 225 Hz. It's so important because A robot cannot move responsively when its controller pauses to imagine the next few frames. LingBot-VA 2.0 predicts during execution, then corrects using each real observation. And it carries only about 13B video parameters while activating roughly 1.9B per token. Bigger robot models usually mean slower reactions, creating a direct conflict between intelligence and control. LingBot-VA 2.0 is trained from scratch for robot control rather than adapted from a video generator built for content creation. Robbyant, an embodied AI company under Ant Group, built it to learn how scenes change under actions, predict what should happen next, and turn those predictions into real-time robot movements. Most video-action systems inherit a tokenizer and video backbone trained mainly to reproduce visual appearance. LingBot-VA 2.0 rebuilds both parts around physical control. Its semantic visual-action tokenizer maps observations toward features from a frozen vision foundation model and learns compact latent actions from frame-to-frame changes using self-supervised inverse and forward dynamics. Unlabeled web video can therefore carry action-relevant training signals without robot action labels. The policy is causal from the start, so every prediction can use only past observations. Its sparse Mixture-of-Experts video backbone has about 13B total parameters, while about 1.9B are active per token, keeping the compute lower during each step. A high-level vision-language planner breaks long tasks into smaller instructions, while the low-level video-action policy handles continuous movement. Foresight Reasoning predicts future visual states while the robot is already acting, then replaces imagined states with every new real observation. Combined with few-step distillation and systems acceleration, the paper reports a peak asynchronous execution frequency of 225 Hz. The model adapts from 10–15 demonstrations, transfers across robot embodiments, and handles some new tasks zero-shot. In the paper’s own evaluations, it reaches 93.6 average on RoboTwin 2.0 and reports stronger real-world results than LingBot-VA and π0.5 across the tested tasks. 🧵 1.

Rohan Paul

11,253 просмотров • 19 дней назад