SimBricks’ Post

Can it run DOOM? Quantum can. Sort of. "Reversibility shenanigans" aside, this is an interesting experiment to try something practical on a QASM simulator (a simulator designed to run on standard devices, modelling quantum logic). Code provided in the article so you can try it yourself. #quantum #quantumcomputing #DOOM https://lnkd.in/g-ut6pw2

This is the new version of Atoms Simulation with new features like better collision handling system and the ability to handle more atoms. https://lnkd.in/dcNHWrr7

Making it easy to get started to configure and run complex simulations has long been a core goal for us with SimBricks, even before our recent adventures towards commercialization. We have had much experience before spending weeks getting up and running with various different simulators. Clearly there has to be a better way, especially for an approach advocating for combining multiple different simulators. Towards this goal, our approach here is to provide known good configurations for each simulator that serve as a starting point to get users up and running immediately, and then allow users to incrementally refine them as needed. In the future we would like to start building up a repository of pre-validated configurations that match specific physical components. That should help manage the validation burden (do it once, then re-use), and increase our users' confidence in their results. For more details, see Marvin's blog post below!

Marvin is blogging about how SimBricks makes it easy for users to simulate even complex systems. Learn more about how we overcome the notorious difficulty and steep learning curves when using simulators, and how we manage the combinatorial complexity of combining multiple simulators. Read more: https://lnkd.in/eprA9czv #FullSystemSimulation #EaseOfUse

Huge strides for the open source world with the latest Flux model by Black Forest Labs 💪 Seems quite competitive with the images from Midjourney and great to see the controlNET canny checkpoint and Lora training with Simple Tuner already out. The dev model size is huge though 😅 Fun test with some images using Runway’s Gen 3 video generation tool. Hugging Face (can test the model here too)- https://lnkd.in/eTJ3u4qX ControlNET - https://lnkd.in/eiR79qFb Lora Training - https://lnkd.in/eRbSYcfU #generativeAI #aiarchitecture #opensource #comfyui #runwayml

Mistral strikes again! 🎉 Let's gooooooo! Yesterday, Mistral released Mixtral 8x22B with a massive 141 billion parameters via magnet link. 🔗🚀 Here's what we know so far: ⚙️ 141 billion parameters with approximately 40 billion active 📜 Context length of 65,000 tokens. (RoPE? - time will tell) 🏗 Base model is fine-tunable 👀 Approximately 260GB VRAM in fp16, 73GB in int4 📜 Apache 2.0 🤗 Available on Hugging Face ⚙️ Similar tokenizer to previous models ❌ No official evaluations or performance numbers ❌ No information about the pertaining dataset ❌ No details on language support Link: https://lnkd.in/degGt2Hb Get ready for another amazing ride with Mistral! 🎢💨 #HackerCulture #MistralAI #OpenSource

Hey everyone! 👋 Have you heard about the latest buzz in the world of computer vision? 🤓 Yolo v8 object detection is taking the industry by storm and making object detection faster and more accurate than ever before. 🚀 Whether you're a developer, researcher or just a curious mind, this is definitely worth checking out. #YOLOv8 #ObjectDetection #ComputerVision #AIMERSOCIETY #AIMERS

I implemented real time object detection using the YOLOv5 algorithm.this is powerful model allows for detecting objects live from webcam #yolov5 #YOLOv5 #openCV #object_detection

Zamba2 a space state 2.7b model (based on mamba2) that doesn’t compromise between inference speed and quality. Pretrained on 3T tokens.. Really impressive evals. https://lnkd.in/d8g_YqV9