Загрузка видео...

Не удалось загрузить видео

На главную

and it works on hardware, inference on a macbook m3 pro. also great at handling highly precise placement / drop thanks to dense patch features solves the task more efficiently that ACT and diffusion.

k7agar's profile picture

atharva ☆

11,969 subscribers

12,785 просмотров • 1 месяц назад •via X (Twitter)

LIVE

Anya Rossi• Live Now

Private livecam show

Комментарии: 0

Нет доступных комментариев

Здесь появятся комментарии из оригинального поста

Похожие видео

Still need to do some cable management, but this is a huge upgrade to my setup. I put my M3 MacBook Air on a SVALT dock (heatsink + fan) and thermal throttling is completely gone. Now when I’m home it basically performs like a MacBook Pro 🔥
0:10

Still need to do some cable management, but this is a huge upgrade to my setup. I put my M3 MacBook Air on a SVALT dock (heatsink + fan) and thermal throttling is completely gone. Now when I’m home it basically performs like a MacBook Pro 🔥

Fabrizio Rinaldi

351,407 просмотров • 2 лет назад

We’re rolling out effort controls in Computer’s model selector. Effort presets combine the orchestrator model and reasoning depth to control how deeply and efficiently Computer works through a task. Available now on web. Coming to mobile and desktop soon.
0:11

We’re rolling out effort controls in Computer’s model selector. Effort presets combine the orchestrator model and reasoning depth to control how deeply and efficiently Computer works through a task. Available now on web. Coming to mobile and desktop soon.

Perplexity

132,675 просмотров • 1 день назад

Julio GUC stays on top of everything with a task list that adapts to the way he works. His system lets you organize tasks by priority, deadlines, and more—great for everything from quick to-dos to big projects →
0:12

Julio GUC stays on top of everything with a task list that adapts to the way he works. His system lets you organize tasks by priority, deadlines, and more—great for everything from quick to-dos to big projects →

Notion

13,732 просмотров • 1 год назад

First Melbourne meal: The famous baked eggs at Hardware Société — ft. chorizo, Spanish cow's milk cheese, spiced concassé, spinach, and more... HIGHLY recommend. Very different spin on breakfast and I enjoyed it
0:19

First Melbourne meal: The famous baked eggs at Hardware Société — ft. chorizo, Spanish cow's milk cheese, spiced concassé, spinach, and more... HIGHLY recommend. Very different spin on breakfast and I enjoyed it

David Lombardi

31,663 просмотров • 11 дней назад

Congrats to everyone at the Presidents Cup — the organizers, the players, the staff, and the volunteers — on a fantastic tournament. Thanks for bringing it back to my hometown, and thanks to the fans for making it such a great time.
0:18

Congrats to everyone at the Presidents Cup — the organizers, the players, the staff, and the volunteers — on a fantastic tournament. Thanks for bringing it back to my hometown, and thanks to the fans for making it such a great time.

Justin Trudeau

220,575 просмотров • 1 год назад

🪄 We're excited to introduce #Lottie by @airbnbdesign support! 🎉 Simply drop a JSON or Lottie file on the patch graph and hook it up to the Lottie Animation patch. 💻 Download the latest version of Origami at:
0:15

🪄 We're excited to introduce #Lottie by @airbnbdesign support! 🎉 Simply drop a JSON or Lottie file on the patch graph and hook it up to the Lottie Animation patch. 💻 Download the latest version of Origami at:

Origami

59,215 просмотров • 3 лет назад

They were building in stealth for 2 years, I was building in stealth for 2 hours… Happy to open source Qwen-2.5-1B-RLCD, 5x faster on-device inference for JSON workloads that need to be type-safe. ⚡️Demo below on a M4 MacBook⚡️ every LLM has the ability to efficiently batch inference every key of a JSON at the same time and generate probabilities from a set of possible categories. No new training required, but it’s easy to optimize if you need! On hugging face now!
0:17

They were building in stealth for 2 years, I was building in stealth for 2 hours… Happy to open source Qwen-2.5-1B-RLCD, 5x faster on-device inference for JSON workloads that need to be type-safe. ⚡️Demo below on a M4 MacBook⚡️ every LLM has the ability to efficiently batch inference every key of a JSON at the same time and generate probabilities from a set of possible categories. No new training required, but it’s easy to optimize if you need! On hugging face now!

Harsha Gundala

1,152,198 просмотров • 2 дней назад

Seedream 5.0 Pro. Unlimited. Now in 2K Every "what if I try it again", at production resolution → Precise editing → Text that stays legible, even in dense layouts → 2K output, ready to post anywhere Try it now on Magnific
0:11

Seedream 5.0 Pro. Unlimited. Now in 2K Every "what if I try it again", at production resolution → Precise editing → Text that stays legible, even in dense layouts → 2K output, ready to post anywhere Try it now on Magnific

Magnific

25,455 просмотров • 2 месяцев назад

Qwen3-TTS Web UI on Mac! Qwen3-TTS web UI now works on Macs too, thanks to contribution from Pierre Bruno, who is also our Pinokio community mod. Thank you! Just install fresh and it should work, I've confirmed this working on my Macbook M1 Max here
0:10

Qwen3-TTS Web UI on Mac! Qwen3-TTS web UI now works on Macs too, thanks to contribution from Pierre Bruno, who is also our Pinokio community mod. Thank you! Just install fresh and it should work, I've confirmed this working on my Macbook M1 Max here

cocktail peanut

50,818 просмотров • 7 месяцев назад

Working on a new topic and hardware for a while: Feet Language Action models. VLAs teach robots to act through vision and language FLAs teach legged robots to act through contact Given a task, terrain, body state, and instruction - decide where, when, and how to make contact
0:18

Working on a new topic and hardware for a while: Feet Language Action models. VLAs teach robots to act through vision and language FLAs teach legged robots to act through contact Given a task, terrain, body state, and instruction - decide where, when, and how to make contact

Irvin (in Japan 🇯🇵)

14,004 просмотров • 2 месяцев назад

Excited to introduce Diffusion Augmented Agents (DAAGs)✨. We give an agent control of a diffusion model, so it can create its own *synthetic experience*.🪄 The result is a lifelong agent that can learn new reward detectors and policies, much more efficiently. Here's how. 👇
0:18

Excited to introduce Diffusion Augmented Agents (DAAGs)✨. We give an agent control of a diffusion model, so it can create its own *synthetic experience*.🪄 The result is a lifelong agent that can learn new reward detectors and policies, much more efficiently. Here's how. 👇

Norman Di Palo

12,931 просмотров • 2 лет назад

Preparing to run the new Llama 3.3 70B on my mac cluster. Downloading model shards onto 3 x M4 Pro Mac Mini and 1 x M3 Max MacBook Pro. AI cluster is connected by Gigabit ethernet switch with EXO Labs
0:18

Preparing to run the new Llama 3.3 70B on my mac cluster. Downloading model shards onto 3 x M4 Pro Mac Mini and 1 x M3 Max MacBook Pro. AI cluster is connected by Gigabit ethernet switch with EXO Labs

Alex Cheema

181,983 просмотров • 1 год назад

🔜 Object Storage on Railway We’re actively working on adding object storage support and we want your feedback: • What would you use it for on Railway? • How are you handling storage today, and what works or doesn’t? • Which features matter most to you?
0:10

🔜 Object Storage on Railway We’re actively working on adding object storage support and we want your feedback: • What would you use it for on Railway? • How are you handling storage today, and what works or doesn’t? • Which features matter most to you?

Railway

54,689 просмотров • 1 год назад

Get Finals-ready! The official 2025 NBA Finals jersey patch is AVAILABLE NOW in-store and online for $20. These limited-edited patches are available while supplies last so act fast if you don’t want to miss out on this piece of Pacers history! Fans coming to our Team Store location can buy a jersey and patch and have it applied to that jersey for no additional charge. We are only able to put the patch on our ’24-25 Nike Jerseys, no exceptions. Approved jerseys are our Icon (Navy), Association (White), Statement (Gold) and City Edition (INDY in White). We will be applying the patch only to jerseys that are bought in-person at our Team Store location at Gainbridge Fieldhouse. If you order a jersey and patch online, the jersey and patch will be shipped to you together for you to apply. If you bought a current jersey from this season and want it applied, you can bring that jersey in-store and purchase a patch and have it applied on-site. Yes 'Cers! Grab yours here:
0:17

Get Finals-ready! The official 2025 NBA Finals jersey patch is AVAILABLE NOW in-store and online for $20. These limited-edited patches are available while supplies last so act fast if you don’t want to miss out on this piece of Pacers history! Fans coming to our Team Store location can buy a jersey and patch and have it applied to that jersey for no additional charge. We are only able to put the patch on our ’24-25 Nike Jerseys, no exceptions. Approved jerseys are our Icon (Navy), Association (White), Statement (Gold) and City Edition (INDY in White). We will be applying the patch only to jerseys that are bought in-person at our Team Store location at Gainbridge Fieldhouse. If you order a jersey and patch online, the jersey and patch will be shipped to you together for you to apply. If you bought a current jersey from this season and want it applied, you can bring that jersey in-store and purchase a patch and have it applied on-site. Yes 'Cers! Grab yours here:

Pacers Team Store

61,014 просмотров • 1 год назад

This Chinese developer launched Llama 70B locally on a MacBook on a plane and for a full 11 hours without internet ran client projects. He was sitting by the window on a transatlantic flight with a MacBook Pro M4 with 64 GB of memory. WiFi on board cost $25 for the flight. He declined. No cloud API, no connection to Anthropic or OpenAI servers, no internet at all. Just a local Llama 3.3 70B on bf16 and his own orchestrator script. The model runs through llama.cpp. Generation speed, 71 tokens per second. Context around 60,000 tokens. Memory usage, 48.6 GiB out of 64. Battery at takeoff, 3 hours 21 minutes. And he gave the orchestrator this system prompt before takeoff: "You are an offline orchestrator running on a single MacBook. There is no network. The only resources you have are local files in /Users/dev/work, the Llama 70B inference server at localhost:8080, and a battery budget of 3 hours 21 minutes. Process the queue at /Users/dev/work/queue.jsonl (one client task per line). For each task: draft → run local evals → save artefact to /Users/dev/work/done/. Save context checkpoints every 12 tasks so you can resume after a battery swap. Stop only on empty queue or when battery drops below 5%." So the system knows exactly what resources it is running on. It knows it has no connection to the outside world for the next 11 hours. It knows it has finite memory and a finite battery. It knows the human will not intervene until the plane lands. The system runs in 1 loop. Takes a task from the queue, runs it through inference, saves the artifact, writes a checkpoint. Task after task, just like that. And only when the battery drops below 5% does the orchestrator automatically pause, waits for the laptop to switch to the backup power bank, and continues from the last checkpoint. Here is what the system actually writes in his log during the flight: "saved context checkpoint 8 of 12 (pos_min = 488, pos_max = 50118, size = 62.813 MiB)" "restored context checkpoint (pos_min = 488, pos_max = 50118)" "prompt processing progress: n_tokens = 50 / 60 818" "task 37016 done | tps = 71 s tokens text → /Users/dev/work/done/proposal_westside.md" Outside the window, clouds, blue sky, and no WiFi. On the tray, 1 MacBook, an open terminal on 2 screens, and an inference server on localhost. From what I have observed, this is the cleanest offline AI workflow I have seen in the past year: 11 hours of flight, $0 for WiFi, and the entire client queue closed before landing.
0:10

This Chinese developer launched Llama 70B locally on a MacBook on a plane and for a full 11 hours without internet ran client projects. He was sitting by the window on a transatlantic flight with a MacBook Pro M4 with 64 GB of memory. WiFi on board cost $25 for the flight. He declined. No cloud API, no connection to Anthropic or OpenAI servers, no internet at all. Just a local Llama 3.3 70B on bf16 and his own orchestrator script. The model runs through llama.cpp. Generation speed, 71 tokens per second. Context around 60,000 tokens. Memory usage, 48.6 GiB out of 64. Battery at takeoff, 3 hours 21 minutes. And he gave the orchestrator this system prompt before takeoff: "You are an offline orchestrator running on a single MacBook. There is no network. The only resources you have are local files in /Users/dev/work, the Llama 70B inference server at localhost:8080, and a battery budget of 3 hours 21 minutes. Process the queue at /Users/dev/work/queue.jsonl (one client task per line). For each task: draft → run local evals → save artefact to /Users/dev/work/done/. Save context checkpoints every 12 tasks so you can resume after a battery swap. Stop only on empty queue or when battery drops below 5%." So the system knows exactly what resources it is running on. It knows it has no connection to the outside world for the next 11 hours. It knows it has finite memory and a finite battery. It knows the human will not intervene until the plane lands. The system runs in 1 loop. Takes a task from the queue, runs it through inference, saves the artifact, writes a checkpoint. Task after task, just like that. And only when the battery drops below 5% does the orchestrator automatically pause, waits for the laptop to switch to the backup power bank, and continues from the last checkpoint. Here is what the system actually writes in his log during the flight: "saved context checkpoint 8 of 12 (pos_min = 488, pos_max = 50118, size = 62.813 MiB)" "restored context checkpoint (pos_min = 488, pos_max = 50118)" "prompt processing progress: n_tokens = 50 / 60 818" "task 37016 done | tps = 71 s tokens text → /Users/dev/work/done/proposal_westside.md" Outside the window, clouds, blue sky, and no WiFi. On the tray, 1 MacBook, an open terminal on 2 screens, and an inference server on localhost. From what I have observed, this is the cleanest offline AI workflow I have seen in the past year: 11 hours of flight, $0 for WiFi, and the entire client queue closed before landing.

Blaze

1,843,280 просмотров • 4 месяцев назад

Palia's got a spring in its step, and our upcoming patch is blooming with new features for you to bask in the great outdoors! 😎🌻 Patch 0.179 🌸 4.23.24 ☀️ Courtyard Add-On 💗 Friendship Level 5 Quests 🎒 Ammo Pouch 💐 More Flowers 🏆 Steam Achievements #Palia #PlayPalia
0:18

Palia's got a spring in its step, and our upcoming patch is blooming with new features for you to bask in the great outdoors! 😎🌻 Patch 0.179 🌸 4.23.24 ☀️ Courtyard Add-On 💗 Friendship Level 5 Quests 🎒 Ammo Pouch 💐 More Flowers 🏆 Steam Achievements #Palia #PlayPalia

Palia

173,495 просмотров • 2 лет назад

Our self-improving agents optimized the full @vLLM_project inference stack, with up to 16% more throughput and interactivity for DeepSeek v4 Pro and Z.ai GLM 5.2 on B200s (no MTP). Every change was verified and our agents got better and faster at it with each iteration.
0:16

Our self-improving agents optimized the full @vLLM_project inference stack, with up to 16% more throughput and interactivity for DeepSeek v4 Pro and Z.ai GLM 5.2 on B200s (no MTP). Every change was verified and our agents got better and faster at it with each iteration.

Asari AI

111,859 просмотров • 1 месяц назад

WOW KIMI K3 ON MY MACBOOK PRO IS MIND BLOWING! I am getting side by side comparisons that I can’t believe. By far the most powerful model I have ever run on a laptop. It is easily better than ChatGPT 3.5 at just a 1 bit quantized and close to ChatGPT 5. It beats most Claude models. More soon!
0:18

WOW KIMI K3 ON MY MACBOOK PRO IS MIND BLOWING! I am getting side by side comparisons that I can’t believe. By far the most powerful model I have ever run on a laptop. It is easily better than ChatGPT 3.5 at just a 1 bit quantized and close to ChatGPT 5. It beats most Claude models. More soon!

Brian Roemmele

107,645 просмотров • 1 месяц назад

Thanks for all the great feedback and comments on Angry, can’t wait for you to hear more!
0:19

Thanks for all the great feedback and comments on Angry, can’t wait for you to hear more!

Mick Jagger

314,229 просмотров • 3 лет назад

I'm happy to announce PhiFlow version 3.0! Just in time for #ICML2024 😀 There are many new features and improvements, such as unstructured meshes, particle processing, improved linear solves and more... Give it a try at:
0:10

I'm happy to announce PhiFlow version 3.0! Just in time for #ICML2024 😀 There are many new features and improvements, such as unstructured meshes, particle processing, improved linear solves and more... Give it a try at:

Thuerey Group at TUM

10,211 просмотров • 2 лет назад