Faris Sweis’ Post

Stable Diffusion 3.0 Launched, Promises Superior Performance. Now, the juicy part... The Upgrade... Stability AI releases Stable Diffusion 3.0 with promises of enhanced image quality and faster performance. A leap forward from its predecessors in efficiency and output. Beyond Images... CEO Emad Mostaque hints at the model's versatility, suggesting future applications could extend to video generation. This opens up new possibilities beyond static images. The Architecture... The new model introduces a novel architecture, designed to be both more efficient and effective. A significant improvement that could redefine the landscape of text-to-image generation. Future Horizons... The release of Stable Diffusion 3.0 not only enhances current capabilities but also paves the way for broader applications. It's a step towards more advanced multimedia generation, setting a new benchmark in the field.

I'm thrilled to announce the publication of my review paper, "Advancements in Generative Modeling: A Comprehensive Survey of GANs and Diffusion Models for Text-to-Image Synthesis and Manipulation," in IEEE Xplore! This paper delves into the two dominant forces in Generative AI (Gen AI): Generative Adversarial Networks (GANs) and Diffusion Models. Here's a glimpse into what you'll find: In-depth exploration of GANs: We delve into the architecture, optimization techniques, and fascinating challenges like mode disintegration and instability. Demystifying Diffusion Models: This section explores how these models, combining noise diffusion and denoising, rewrite the generative narrative. We discuss their high-fidelity photo generation, precise distribution coverage, and scalability advantages. The Power of Collaboration: The paper sheds light on the interaction between GANs and Diffusion Models, highlighting scenarios where each excels and how their collaboration unlocks even more powerful capabilities. A Roadmap for the Future: We conclude by identifying key areas ripe for exploration and development in the ever-evolving landscape of generative modeling. This paper serves as a comprehensive resource for researchers and practitioners who want to navigate the exciting world of GANs and Diffusion models. Feel free to download the paper here https://lnkd.in/gHg5wp4D and reach out if you have any questions! Primary Author: PRIYANSHU DESHMUKH Author 2: Pranav Ambulkar Author 3: Pranoti Sarjoshi Author 4: Harshal Dabhade Author 5 and guide: Mr. Saurabh Shah CSE PRMCEAM, Badnera Krupa Ravi Chi Wang Flo Crivello #generativemodeling #GANs #diffusionmodels #texttoimage #machinelearning #AI #research #IEEE #Xplore

Unveiling the Power of Transformation: Dive into the World of Transformer Architecture! 🌐✨ Discover how this groundbreaking model is reshaping AI, offering unparalleled insights and capabilities. I present a small self-written summary of the research paper 'Attention is all you need' for those who are still unaware of the transformer architecture. Link of the paper: https://lnkd.in/eu7t9n9m Any feedback will be much appreciated 😊 #AIRevolution #TransformerArchitecture #FutureIsNow

What if intelligent agents could live and experience a built environment before it was construction? In a recent collaboration, researchers from Stanford University and Google DeepMind introduced generative agents—intelligent entities that utilise large language models to simulate human behaviour, enabling them to remember, reflect, and engage within an environment. These agents cook, work and engage in dialog as they live from day to day. In a village of 25, these agents ended up making friends, asking each other out of dates and even planning a party together. As Architects and Engineers, we employ simulation methods to understand human movement and operations within complex buildings and civic spaces. The introduction of generative agents provide opportunity to conduct social simulations during initial design iterations, offering insights into how individuals might inhabit and interact within our designs. Huge potential in this space, amazing work Joon Sung Park, Joseph C. O'Brien, Carrie J. Cai, Meredith Morris, Percy Liang and Michael Bernstein FYI Matt Goldsberry Miguel Rus M. Hank Haeusler Lin (Leo) Meng Sean M. Carr Alex Leiva Alex Leese John Vollugi Laurent Deleu Clayton Riddle Please read the fascinating full paper can be found here: https://lnkd.in/gy-SKyWQ Image referenced from the paper above. #generativeai #artificialintelligence #inspiration #architecture #computationaldesign #generativedesign #ai #simulations #largelanguagemodel #largelanguagemodels #architecturedesign #design #gamification

Thanks for sharing this Jeremy Graham. I was a big fan of Westworld and am glad (and scared) to see it gradually becoming true (perhaps in the next decade?). Recursion + generative AI has infinite possibilities (AutoGPT alike). While unleashing its massive potential for productivity, we also need to be very cautious of what human tasks we are delegating to the AI. Second note: LLM models are largely criticised for their lack of ability for deductive reasoning and mathematical operation, and thus lack of spatial understanding. This work seems promising in breaking this perception in the future: https://lnkd.in/gKF-pnMr

Want to learn about the recent advancements in GANs in one place? I highly recommend reading this paper! Congratulations PRIYANSHU DESHMUKH on your amazing work and publishing the paper in IEEE! #ai #data #gan

𝗧𝗵𝗶𝘀 𝗶𝘀 𝘄𝗶𝗹𝗱! 𝗖𝗵𝗶𝗻𝗮 𝗷𝘂𝘀𝘁 𝗮𝗻𝗻𝗼𝘂𝗻𝗰𝗲𝗱 𝗶𝘁𝘀 𝗿𝗶𝘃𝗮𝗹 𝘁𝗼 𝗢𝗽𝗲𝗻𝗔𝗜’𝘀 𝗦𝗼𝗿𝗮 😳 Like Sora, Vidu is capable of producing 16-second clips at 1080p resolution. Vidu is based on a Universal Vision Transformer (U-ViT) architecture This allows it to simulate the real physical world with multi-camera view generation. The crazy part? The architecture of this model was reportedly initiated in September 2022. As such it would predate the diffusion transformer (DiT) architecture used by Sora 🤯 Seeing is no longer believing. Wild. P.S. '𝗩𝗶𝗲𝘄 𝗺𝘆 𝗯𝗹𝗼𝗴' for your daily briefing on Generative AI

Lately, I have been more and more intrigued by the fascinating realm of artificial intelligence (AI) and its evolving relationship with architecture! In my latest blog post, I've expressed my opinion on the intersection of AI and architecture, and how this dynamic duo has the potential to transform the creative landscape. I believe that AI is not here to steal our spotlight but to complement and amplify the creative process we architects hold dear. As we navigate this digital frontier, envisioning a future where AI becomes a powerful ally in our design endeavors, I can't wait to share my insights and perspectives with you. Below is the link to my article on Medium.✨ If you are not a Medium member, use the following link to get full access to the article - https://lnkd.in/dUMq2Hnm #architecture #ai #software #future #career

Is AI God? One of the most compelling questions I grappled at this years Midwest Architecture Community Collaboration event. Thank you Youssef Haddad for Chairing this great event! The theme for this year was #artificialintelligence (surprise?). The event ended with a great panel Q&A featuring 2 members of #CAF(Chief Architect Forum)- E.G. Nadhan and Grant Ecker! Both Nadhan and Grant were speakers too. Here are some other good questions about AI to ponder (post your thoughts in the comments below): 1: Is AI God? 2: How can AI be used to optimize IT systems? 3: How can AI be used to transform old legacy systems to more modern systems? 4: Are humans living in a simulation (cue the matrix scene)? #genai #artificialintelligenceforbusiness #enterprisearchitecture #businessarchitecture