正在加载视频...

视频加载失败

加载此视频时出现问题。这可能是由于临时网络问题，或视频可能不可用。

i trained Llama 2 on a $5 computer with less processing power than a thermostat. it’s called 🥧 PiTorch: an ML library for Raspberry Pi Zeros. i cover how to make it from scratch, from writing assembly GPU kernels to sending bytes over wires. (no hardware background needed!)

Mason Wang

3,961 subscribers

130,131 次观看 • 5 个月前 •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

0 条评论

暂无评论

原始帖子的评论将显示在这里

相关视频

My course is finally OUT! I spent over a year on Creating Creatures for Games and I honestly couldn’t be prouder of the result. It covers how to build this kaiju from scratch, with all steps shown: sculpting, retopo, baking, texturing and unreal lighting. This shot is from Unreal with a bit of post processing on it Anim and rig by Kiel Figgins Shot lighting by Elpida Kyriakou Audio by BloodYuki

My course is finally OUT! I spent over a year on Creating Creatures for Games and I honestly couldn’t be prouder of the result. It covers how to build this kaiju from scratch, with all steps shown: sculpting, retopo, baking, texturing and unreal lighting. This shot is from Unreal with a bit of post processing on it Anim and rig by Kiel Figgins Shot lighting by Elpida Kyriakou Audio by BloodYuki

Henning Sanden

38,306 次观看 • 2 个月前

i've been paying 200$/month for AI tools for a while now claude, chatgpt, all of it thought that was just the price of doing things properly then i read this and had to close my laptop for a second someone bought a used GPU from eBay for $700 five years old. RTX 3090. the kind of card people sell when they upgrade for gaming plugged in a free model called Qwen 3.6 that alibaba quietly dropped last month and it scored 84.1 on vision tasks claude 4.5 opus scored 77.0 the 200$/month subscription lost to a free model running on old hardware the math gets worse after 4 months you've paid off the GPU from month 5 onwards you pay $8 for electricity that's it no subscription. no per message cost. nothing leaves your computer. ever. turns out i didn't have to is anyone else just finding out about this or was i the last one to know?

i've been paying 200$/month for AI tools for a while now claude, chatgpt, all of it thought that was just the price of doing things properly then i read this and had to close my laptop for a second someone bought a used GPU from eBay for $700 five years old. RTX 3090. the kind of card people sell when they upgrade for gaming plugged in a free model called Qwen 3.6 that alibaba quietly dropped last month and it scored 84.1 on vision tasks claude 4.5 opus scored 77.0 the 200$/month subscription lost to a free model running on old hardware the math gets worse after 4 months you've paid off the GPU from month 5 onwards you pay $8 for electricity that's it no subscription. no per message cost. nothing leaves your computer. ever. turns out i didn't have to is anyone else just finding out about this or was i the last one to know?

Crypto Mavka

14,126 次观看 • 3 个月前

China's government built this with 1 trillion dollars. It now has more bullet trains than the rest of the world combined. They didn't build it for profit, but to provide a public service. Here, our government is experimenting with everything just to figure out How to make a profit from it. How to collect tolls from roads. How to collect tax from every new item. And yes, I have seen them spending millions on airports that no one uses.

China's government built this with 1 trillion dollars. It now has more bullet trains than the rest of the world combined. They didn't build it for profit, but to provide a public service. Here, our government is experimenting with everything just to figure out How to make a profit from it. How to collect tolls from roads. How to collect tax from every new item. And yes, I have seen them spending millions on airports that no one uses.

🚨Indian Gems

88,520 次观看 • 4 个月前

Perplexity Computer just opened a PR on my repo, starting with a simple task... improving SEO What stands out is how it executes. Exploring with Gemini, then switching to Opus when it’s time to actually write code (I haven’t seen it use Codex or Grok) I’m going to ramp up the complexity from here. So far, it’s delivered on every task. I have a few more in progress that I’ll be sharing over the next few days.

Perplexity Computer just opened a PR on my repo, starting with a simple task... improving SEO What stands out is how it executes. Exploring with Gemini, then switching to Opus when it’s time to actually write code (I haven’t seen it use Codex or Grok) I’m going to ramp up the complexity from here. So far, it’s delivered on every task. I have a few more in progress that I’ll be sharing over the next few days.

Daniel San

15,317 次观看 • 6 个月前

joshua mentioned cookscoups! #SCOUPS Q. any food you want to try if you went on cookscoups? 🦌 german sausages! but he has to make it from scratch, mince up the meat, pipe it into the casing and let it rest for 3 days in the fridge. i want him to bring it over and cook it on a charcoal grill. i like the smoky flavour.

joshua mentioned cookscoups! #SCOUPS Q. any food you want to try if you went on cookscoups? 🦌 german sausages! but he has to make it from scratch, mince up the meat, pipe it into the casing and let it rest for 3 days in the fridge. i want him to bring it over and cook it on a charcoal grill. i like the smoky flavour.

ss.

135,282 次观看 • 8 个月前

There's an open-source home robot vacuum you can actually build yourself. It's called OOMWOO. Runs on a Raspberry Pi with 2D LiDAR mapping, ROS2 navigation, and native Home Assistant integration. No cloud dependency, no vendor lock-in. The team spent real time researching before designing it. They reviewed the entire 2025-2026 robot vacuum market, from budget to flagship, and found real-world cleaning doesn't track advertised suction power. The anti-tangle brush design alone is worth mentioning. A tapered rubber roller that resists hair-wrap, one of the most common complaints with commercial models, easy to 3D-print yourself. If you've got a Raspberry Pi and a 3D printer sitting around, this is a genuinely interesting weekend project.

There's an open-source home robot vacuum you can actually build yourself. It's called OOMWOO. Runs on a Raspberry Pi with 2D LiDAR mapping, ROS2 navigation, and native Home Assistant integration. No cloud dependency, no vendor lock-in. The team spent real time researching before designing it. They reviewed the entire 2025-2026 robot vacuum market, from budget to flagship, and found real-world cleaning doesn't track advertised suction power. The anti-tangle brush design alone is worth mentioning. A tapered rubber roller that resists hair-wrap, one of the most common complaints with commercial models, easy to 3D-print yourself. If you've got a Raspberry Pi and a 3D printer sitting around, this is a genuinely interesting weekend project.

Oliver Prompts

585,261 次观看 • 1 个月前

I just sequenced a human genome to 30× coverage entirely at home. As far as I know, this is the first time this has been done. I didn’t step foot in a lab once. Every step - from saliva collection, to running the sequencer - took place in a single room with a dining table + kitchenette. Six weeks ago, I had never done wet lab biology before. I used an Oxford Nanopore P2 Solo - the only commercially available sequencing device portable enough to do 30x human genome sequencing at home. Biggest takeaway - I could build something that combined software, hardware, and molecular biology far faster than I thought was possible. I can name >100 specific instances where AI helped me solve a technical problem that would previously have blocked me because I lacked access to a domain expert. For example: how do I save my sequencing run when my DNA extraction yield is 4x lower than I need it to be, and I have this limited set of reagents to hand? To make this work, I had to navigate multiple disciplines: - writing software to monitor sequencing runs and orchestrate remote GPU infra for basecalling - learning + executing 5 hour long molecular biology protocols - building a hardware device to quantify DNA concentration Apologies for the hyperbole, but I feel super lucky to be living in 2026. A few weeks ago I decided to sequence a human genome to 30x at home. Then I actually did it. And I did it really quickly.

I just sequenced a human genome to 30× coverage entirely at home. As far as I know, this is the first time this has been done. I didn’t step foot in a lab once. Every step - from saliva collection, to running the sequencer - took place in a single room with a dining table + kitchenette. Six weeks ago, I had never done wet lab biology before. I used an Oxford Nanopore P2 Solo - the only commercially available sequencing device portable enough to do 30x human genome sequencing at home. Biggest takeaway - I could build something that combined software, hardware, and molecular biology far faster than I thought was possible. I can name >100 specific instances where AI helped me solve a technical problem that would previously have blocked me because I lacked access to a domain expert. For example: how do I save my sequencing run when my DNA extraction yield is 4x lower than I need it to be, and I have this limited set of reagents to hand? To make this work, I had to navigate multiple disciplines: - writing software to monitor sequencing runs and orchestrate remote GPU infra for basecalling - learning + executing 5 hour long molecular biology protocols - building a hardware device to quantify DNA concentration Apologies for the hyperbole, but I feel super lucky to be living in 2026. A few weeks ago I decided to sequence a human genome to 30x at home. Then I actually did it. And I did it really quickly.

Seth Howes

412,115 次观看 • 3 个月前

I timed a 100m race from the tv broadcast. No timing hardware, no finish sensors, nothing trained. All open source vision libraries. we called the winner at 11.26s. the official time was 11.23. the stack: - trackers (BoT-SORT + camera motion compensation) to hold all 8 runners through the pan - RF-DETR for the runner detections - opencv to find the finish line, because it's just white paint on the track - fft magic to hear the starting gun on the soundtrack tell me if you want the tutorial on how i built it!

I timed a 100m race from the tv broadcast. No timing hardware, no finish sensors, nothing trained. All open source vision libraries. we called the winner at 11.26s. the official time was 11.23. the stack: - trackers (BoT-SORT + camera motion compensation) to hold all 8 runners through the pan - RF-DETR for the runner detections - opencv to find the finish line, because it's just white paint on the track - fft magic to hear the starting gun on the soundtrack tell me if you want the tutorial on how i built it!

Alex Bodner

98,817 次观看 • 21 天前

I explored a further possibility with local models: Qwen3.6 35B A3B + NVIDIA LocateAnything-3B as a local Computer Use agent (proof of concept). In the demo, I asked it to switch my Mac to light mode. It did. Then back to dark. Did that too — finding the right toggle in System Settings, clicking it, and verifying the change itself. It's fully screenshot-based, so no Accessibility API needed. If it's on screen, the agent can see it and act on it. This runs entirely on your own hardware — private, local, built from two small open models.

I explored a further possibility with local models: Qwen3.6 35B A3B + NVIDIA LocateAnything-3B as a local Computer Use agent (proof of concept). In the demo, I asked it to switch my Mac to light mode. It did. Then back to dark. Did that too — finding the right toggle in System Settings, clicking it, and verifying the change itself. It's fully screenshot-based, so no Accessibility API needed. If it's on screen, the agent can see it and act on it. This runs entirely on your own hardware — private, local, built from two small open models.

stevibe

44,231 次观看 • 3 个月前

Be the detail you want to see in the world Custom hybrid UIKit gestures + SwiftUI for all the actual content - both reading/writing from a central observable variable store (not using vermont but a custom library I wrote that’s based off of simple closures and AsyncStream) for completely synced animations. Will share more publicly soon on how it’s built :-)

Be the detail you want to see in the world Custom hybrid UIKit gestures + SwiftUI for all the actual content - both reading/writing from a central observable variable store (not using vermont but a custom library I wrote that’s based off of simple closures and AsyncStream) for completely synced animations. Will share more publicly soon on how it’s built :-)

Tyler Angert

16,760 次观看 • 7 个月前

This isn’t a 3D model thrown in a box. This is pure code. No Blender, no Maya, no shortcuts. Every voxel, every jump, every movement scripted from scratch. Generative art at this level means thinking like a designer, a coder, and an artist all at once. It’s not about placing pixels. It’s about writing the rules of how a frog should exist in the Bitcoin chain. BLOCKFROGS isn’t just generative. It’s engineered. Be part of this experience Wl = 2 Freemints 5 WL to snatch Post or Quote about FrogsHood Drop the link

This isn’t a 3D model thrown in a box. This is pure code. No Blender, no Maya, no shortcuts. Every voxel, every jump, every movement scripted from scratch. Generative art at this level means thinking like a designer, a coder, and an artist all at once. It’s not about placing pixels. It’s about writing the rules of how a frog should exist in the Bitcoin chain. BLOCKFROGS isn’t just generative. It’s engineered. Be part of this experience Wl = 2 Freemints 5 WL to snatch Post or Quote about FrogsHood Drop the link

FrogsHood

16,264 次观看 • 1 年前

Vercel and Netlify charge $20/seat a month. Heroku killed its free tier. Go over your bandwidth and you get surprise bills, sometimes thousands. There's an open-source alternative to all three. For $0. It's called Coolify. You install it on any server you own, a $5 VPS, a Raspberry Pi, an old laptop, anything with SSH. Then you deploy everything from one dashboard. > Static sites, full-stack apps, databases, and APIs > 280+ one-click services (WordPress, Ghost, n8n, Supabase, and more) > Free auto-renewing SSL and automatic S3 backups > Pull request preview deployments > Push-to-deploy from GitHub, GitLab, and Bitbucket No vendor lock-in. Your configs live on your server, so if you stop using Coolify, everything still runs. A 5-person team on Vercel pays $100-500/month. The same team on a $5 VPS with Coolify pays $5. Total. 100% Free. Open Source.

Vercel and Netlify charge $20/seat a month. Heroku killed its free tier. Go over your bandwidth and you get surprise bills, sometimes thousands. There's an open-source alternative to all three. For $0. It's called Coolify. You install it on any server you own, a $5 VPS, a Raspberry Pi, an old laptop, anything with SSH. Then you deploy everything from one dashboard. > Static sites, full-stack apps, databases, and APIs > 280+ one-click services (WordPress, Ghost, n8n, Supabase, and more) > Free auto-renewing SSL and automatic S3 backups > Pull request preview deployments > Push-to-deploy from GitHub, GitLab, and Bitbucket No vendor lock-in. Your configs live on your server, so if you stop using Coolify, everything still runs. A 5-person team on Vercel pays $100-500/month. The same team on a $5 VPS with Coolify pays $5. Total. 100% Free. Open Source.

Simplifying AI

14,549 次观看 • 1 个月前

Free NVIDIA GPU with 16 GB VRAM GPU for Running Local LLMs! If you want to master local LLMs but you're waiting until you can afford a $1,500 GPU, you're honestly not going to make it. The open source AI ecosystem is moving way too fast for you to wait on your budget to catch up. Especially when you can build a bleeding edge inference engine from scratch right now, completely for free. You don't need a heavy local rig to start. Google is literally letting you use an enterprise grade NVIDIA Tesla T4 GPU for $0/hour. At standard cloud computing rates (~$0.20/hr), Google Colab’s 4 hour daily free tier hands you roughly $24 worth of data center tier GPU compute every single month. And most people just waste it. Let’s talk about the hardware you get access to for free. The NVIDIA Tesla T4 is an absolute workhorse: - Architecture: NVIDIA Turing (TU104) - VRAM: 16GB GDDR6 (320 GB/s bandwidth) - Compute: 320 Tensor Cores | 2560 CUDA Cores - Performance: 130 TOPS INT8 | 8.1 TFLOPS FP32 - Power: Sipping energy at a max 70W TDP This is the exact same hardware I used to run DeepMind's Gemma 4 26B A4B QAT MoE at a 250,000 context window without a single Out Of Memory (OOM) crash. If you have a web browser and 10 minutes, you have everything you need. I’ve put together a fully documented, cell by cell Google Colab notebook that teaches you exactly how to do this. Here is what the notebook actually teaches you: - How to provision an Ubuntu Linux environment with CUDA 13.0 and verify your driver stack. - How to pull the source code and compile the latest llama.cpp C++ binaries from scratch, specifically optimizing the build for your exact GPU using the -DCMAKE_CUDA_ARCHITECTURES=native flag. - How to directly download quantized local LLMs (GGUF format) straight from HuggingFace using the CLI. - How to manage 16GB VRAM limits, offload neural network layers to the GPU, and push massive context windows. Compile raw llama.cpp, ollama run a model, or spin up the LM Studio CLI. Pick whatever stack you are comfortable with. just start building. No hardware. No credit card. No excuses. Bookmark this post right now so you don't lose the tutorial. Even if you don't have time to run it today, you are going to want this workflow in your engineering toolkit. The link to the free Colab Notebook is in the comments below. Lemme know if you need more tutorials like this.

Free NVIDIA GPU with 16 GB VRAM GPU for Running Local LLMs! If you want to master local LLMs but you're waiting until you can afford a $1,500 GPU, you're honestly not going to make it. The open source AI ecosystem is moving way too fast for you to wait on your budget to catch up. Especially when you can build a bleeding edge inference engine from scratch right now, completely for free. You don't need a heavy local rig to start. Google is literally letting you use an enterprise grade NVIDIA Tesla T4 GPU for $0/hour. At standard cloud computing rates (~$0.20/hr), Google Colab’s 4 hour daily free tier hands you roughly $24 worth of data center tier GPU compute every single month. And most people just waste it. Let’s talk about the hardware you get access to for free. The NVIDIA Tesla T4 is an absolute workhorse: - Architecture: NVIDIA Turing (TU104) - VRAM: 16GB GDDR6 (320 GB/s bandwidth) - Compute: 320 Tensor Cores | 2560 CUDA Cores - Performance: 130 TOPS INT8 | 8.1 TFLOPS FP32 - Power: Sipping energy at a max 70W TDP This is the exact same hardware I used to run DeepMind's Gemma 4 26B A4B QAT MoE at a 250,000 context window without a single Out Of Memory (OOM) crash. If you have a web browser and 10 minutes, you have everything you need. I’ve put together a fully documented, cell by cell Google Colab notebook that teaches you exactly how to do this. Here is what the notebook actually teaches you: - How to provision an Ubuntu Linux environment with CUDA 13.0 and verify your driver stack. - How to pull the source code and compile the latest llama.cpp C++ binaries from scratch, specifically optimizing the build for your exact GPU using the -DCMAKE_CUDA_ARCHITECTURES=native flag. - How to directly download quantized local LLMs (GGUF format) straight from HuggingFace using the CLI. - How to manage 16GB VRAM limits, offload neural network layers to the GPU, and push massive context windows. Compile raw llama.cpp, ollama run a model, or spin up the LM Studio CLI. Pick whatever stack you are comfortable with. just start building. No hardware. No credit card. No excuses. Bookmark this post right now so you don't lose the tutorial. Even if you don't have time to run it today, you are going to want this workflow in your engineering toolkit. The link to the free Colab Notebook is in the comments below. Lemme know if you need more tutorials like this.

Alok

182,483 次观看 • 2 个月前

Last night at the casino hotel in Miami, I invited this girl over to my suite. We’re chilling, I open my laptop and start gambling on Rainbet. She tells me her Uber Black cost $200 and asks if I can send it in crypto. I go to withdraw $200 while she’s sitting next to me. so I go to withdraw on rainbet & as my cursor’s hovering over “withdraw,” she taps two extra zeros in less than half a second before I click. $20,000 gone straight to her wallet. She instantly runs for the door. I call hotel security they caught her before she made it out the building. I ended up not getting the money back and they didn’t even arrest her because apparently police don’t know how to read a fucking solscan. So $20k in SOL gone & all she got was a lifetime ban from the casino.

Sensitive content

Last night at the casino hotel in Miami, I invited this girl over to my suite. We’re chilling, I open my laptop and start gambling on Rainbet. She tells me her Uber Black cost $200 and asks if I can send it in crypto. I go to withdraw $200 while she’s sitting next to me. so I go to withdraw on rainbet & as my cursor’s hovering over “withdraw,” she taps two extra zeros in less than half a second before I click. $20,000 gone straight to her wallet. She instantly runs for the door. I call hotel security they caught her before she made it out the building. I ended up not getting the money back and they didn’t even arrest her because apparently police don’t know how to read a fucking solscan. So $20k in SOL gone & all she got was a lifetime ban from the casino.

29

79,732 次观看 • 11 个月前

My Lagos Liga folks called me, told me they were trying to close a deal with a watch manufacturer based in California and they needed someone would could get the deal over the line. I spent 7’ on the phone with Flexaago. 5’ was spent on sharing personal stories. 2 minutes was spent on agreeing the number for Lagos Liga. Done deal. This is what I do. On to the next one.

My Lagos Liga folks called me, told me they were trying to close a deal with a watch manufacturer based in California and they needed someone would could get the deal over the line. I spent 7’ on the phone with Flexaago. 5’ was spent on sharing personal stories. 2 minutes was spent on agreeing the number for Lagos Liga. Done deal. This is what I do. On to the next one.

JJ. Omojuwa

92,759 次观看 • 1 年前

for some brands from the L’Oréal group, I created this prompt to easily generate an isometric office diorama that can be adapted to any brand and material it works with a simple Nano Banana 2 prompt and is easy to animate with Kling 3.0 this is how I imagine my future office

INK

40,206 次观看 • 6 个月前

Last night at the casino hotel in Miami, I invited this girl over to my suite. We're chilling, I open my laptop and start gambling on Rainbet. She tells me her Uber Black cost $200 and asks if I can send it in crypto. I go to withdraw $200 while she's sitting next to me. so | go to withdraw on rainbet & as my cursor's hovering over "withdraw," she taps two extra zeros in less than half a second before I click. $20,000 gone straight to her wallet. She instantly runs for the door. I call hotel security they caught her before she made it out the building. I ended up not getting the money back and they didn't even arrest her because apparently police don't know how to read a fucking solscan. So $20k in SOL gone & all she got was a lifetime ban from the casino.

Last night at the casino hotel in Miami, I invited this girl over to my suite. We're chilling, I open my laptop and start gambling on Rainbet. She tells me her Uber Black cost $200 and asks if I can send it in crypto. I go to withdraw $200 while she's sitting next to me. so | go to withdraw on rainbet & as my cursor's hovering over "withdraw," she taps two extra zeros in less than half a second before I click. $20,000 gone straight to her wallet. She instantly runs for the door. I call hotel security they caught her before she made it out the building. I ended up not getting the money back and they didn't even arrest her because apparently police don't know how to read a fucking solscan. So $20k in SOL gone & all she got was a lifetime ban from the casino.

29

67,682 次观看 • 5 个月前

Someone built an AI you power with a hand crank it is called CrankGPT and its running without any battery, internet or data centre just you turning a handle like it is 1900 SqueezLabs built it using a Raspberry Pi 5 with 8GB of RAM an audio card and a 20 watt hand crank generator it takes about 30 seconds of cranking to boot into a working voice assistant and the onboard capacitor gives you roughly 20 seconds of runtime before you have to start cranking again they have used it to generate small images even write code 🤯 this proves you can run a real AI model on almost no power while everyone is building billion dollar data centres two guys put AI in a box

Someone built an AI you power with a hand crank it is called CrankGPT and its running without any battery, internet or data centre just you turning a handle like it is 1900 SqueezLabs built it using a Raspberry Pi 5 with 8GB of RAM an audio card and a 20 watt hand crank generator it takes about 30 seconds of cranking to boot into a working voice assistant and the onboard capacitor gives you roughly 20 seconds of runtime before you have to start cranking again they have used it to generate small images even write code 🤯 this proves you can run a real AI model on almost no power while everyone is building billion dollar data centres two guys put AI in a box

Sweep

11,610 次观看 • 2 个月前