Cohorte

Hello world 🖖 We’re living exciting times, isn’t it? We can work from everywhere, learn anything, collaborate with anybody, and create almost anything on the web! There are limitless possibilities and every day many people are changing their lives and adapting their work to their lifestyle - not the reverse. This is particularly true for tech and data workers. You need to access data, and a machine terminal and that’s it 🚀 At Cohorte, we believe in a future where people & businesses collaborate and build together with high agility and minimal friction. A new order where intermediaries and rigid organizations give way to communities of independent and passionate talents to self-organize and deliver their best work every day while finding meaning and fulfillment. In 2022, companies need access to on-demand expertise from the best talents worldwide to build fast, shift quickly, and adjust to production demands. Technology is becoming an essential component of any business, and having the right mix of skills is critical for success. That's why we want to create a platform where skilled and passionate people with different technology skills can organize themselves to collaborate and complete complex projects. We believe that between freelance consulting catalogs and traditional consulting firms, there's a place for an alternative order, where people can be independent, enjoy flexibility, join a structured network and expand their business. In such an order, there is no room for subordination, but only for supportive and mentoring relationships, where people grow together, and experienced talents help younger ones to succeed. On the one hand, independent workers enjoy flexibility and freedom without sacrificing learning from peers, the scalability of their work, and the scope of their missions. On the other hand, businesses benefit from the agility and premium services from highly committed and cohesive professionals. ⚡️Do you believe in our project? Click the “Follow” button to support us. ⚡️If you are a tech/data freelancer, join our network, get access to our opportunities and discover our capabilities to scale and expand your business. ⚡️If you are a business, follow us, and boost your projects now by onboarding the best tech and data collectives with the right combination of skills. Let’s shape the new normal 🙌 #technology #data #consulting

Sharing the Language of Imagination! The world of image generation with Midjourney just got a whole lot more collaborative! Introducing "style codes": a revolutionary way to share and explore unique visual styles. These codes act like a secret language, allowing users to communicate specific artistic preferences and generate stunning results. Imagine you have a vision for a "futuristic city made of glass and smoke." But how do you translate that complex idea into an image? Enter style codes! By utilizing specific codes like 1053330878, 2813629345, 4244977961, and 1558173677, you can unlock a range of unique interpretations. Each code will subtly tweak the artistic approach, resulting in a diverse set of captivating cityscapes. Here's how it works: - Style codes capture the essence of a specific artistic style. This can encompass elements like lighting, brushstrokes, color palettes, and overall mood. - Sharing these codes opens doors to collaboration. Artists can share their signature styles, allowing others to explore them and build upon them. - The possibilities are endless! Experiment with different codes, combine them and see what unique visions come to life. The Benefits of Style Codes: - Effortlessly achieve desired aesthetics. No more struggling to describe your dream image – use style codes for instant results. - Discover new artistic directions. Explore a vast library of styles and find inspiration for your next project. - Fuel creative collaboration. Share your style codes and inspire others, while learning from theirs. - Midjourney's style codes are a game-changer! They empower creators to share their artistic language and unlock entirely new visual landscapes. As this technology evolves, we can expect even more exciting possibilities for collaborative image generation. _____________ ✔️ Click "Follow" on the Cohorte page for daily AI engineering news. Credits: Ethan Mollick

🦓 YOUDREAM: Crafting Imaginary Creatures with Anatomical Precision! 🦄 Researchers from the University of Texas at Austin and the University of Massachusetts Amherst have unveiled YOUDREAM, a new method that takes text-to-3D generation to the next level by enabling anatomically controllable animal creation. 🦴 YOUDREAM stands out by: - Going beyond text and image prompts: Offers users the ability to guide generation using 3D pose priors, unlocking creativity for unreal and mythical creatures. - Preserving anatomical consistency: Generates 3D animals that are biologically plausible and avoid the common pitfalls of other text-to-3D methods (Janus-head problem, deformed limbs, etc.). - Automating pose generation: Utilizes a multi-agent LLM system to adapt poses from a library of existing animal skeletons, making it easy to create new creatures. - Providing a user-friendly pose editor: Offers an intuitive tool for artists to customize and refine poses for even greater control over the generated animals. YOUDREAM excels in several aspects: - Creating mythical and unreal creatures: Generates high-quality 3D models of animals that are not found in nature or cannot be easily described with text alone. - Generating anatomically correct animals: Produces models with realistic bone structures and proportions, ensuring biological plausibility. Improving user control: Offers a new level of creative freedom for artists and 3D modelers. - Outperforming existing text-to-3D methods: User studies show a strong preference for YOUDREAM-generated animals in terms of naturalness and text-image alignment. YOUDREAM opens up a world of possibilities for creating unique and imaginative 3D animal models, pushing the boundaries of creative expression in AI-powered content generation. Read the full paper: https://lnkd.in/d29WT78A #YOUDREAM #TextTo3D #AnimalGeneration #AnatomicalControl #AI #DeepLearning ✔️ Click "Follow" on the Cohorte page for daily AI engineering news. Credits: Sandeep Mishra, Oindrila Saha, Alan C. Bovik

AI Actors, The Next Big Thing in Entertainment, or a Dystopian Dream? The world of entertainment is on the cusp of a revolution. Artificial intelligence (AI) is now capable of generating incredibly realistic human characters that never existed. This opens a treasure trove of possibilities for filmmakers and advertisers alike. Imagine: - Creating the perfect character for a specific role: Shy teenager? Grizzled war veteran? AI can craft a character with meticulous detail, from their appearance to their mannerisms. - Breaking down language barriers: AI-generated actors can speak any language flawlessly, opening doors to global audiences. - Reducing production costs: No need for expensive sets, costumes, or travel – AI actors can be filmed entirely in a digital environment. But here's the rub: - The "Uncanny Valley": While AI-generated humans are becoming increasingly realistic, there's still a feeling of "off-ness" that can be unsettling. Can AI ever truly capture the nuances of human emotion? - The Ethics of Deepfakes: AI-generated actors raise concerns about deepfakes, where real people's faces are superimposed on others for malicious purposes. - The Human Touch: Will AI ever replace the raw talent and emotional depth of real actors? Is there a risk of homogenization in entertainment? - The Future is Unclear: Some believe AI actors will eventually become indistinguishable from humans on screen. At that point, does it even matter if they're real? This sparks a fascinating conversation: Do audiences crave authenticity, or will AI-generated characters be just as compelling? How can we ensure the ethical use of AI in entertainment? How will the role of human actors evolve alongside this technology? This isn't just about technology – it's about the future of storytelling. AI actors offer exciting possibilities, but they also raise important questions. As we move forward, it's crucial to approach this technology with a critical eye, ensuring it enhances, rather than diminishes, the magic of entertainment. What are your thoughts on AI actors? Share your opinion in the comments below! _____________ ✔️ Click "Follow" on the Cohorte page for daily AI engineering news. 

🐙 Octo-planner: Bringing the Power of Planning Agents to Your Devices! 📱 Researchers at Nexa AI and MIT have developed Octo-planner, an efficient on-device planning agent that enables AI-powered assistance on resource-constrained devices like smartphones. 🧠 Octo-planner tackles the challenges of on-device AI by: - Utilizing a Planner-Action framework: Separates planning and action execution into two distinct components, allowing for specialized optimization and modularity. - Prioritizing fine-tuning over in-context learning: Reduces computational cost and memory requirements, making it ideal for edge devices with limited resources. - Leveraging GPT-4 for data generation and validation: Creates a high-quality dataset for fine-tuning the planner agent, ensuring accurate and diverse planning capabilities. - Employing a multi-LoRA training method: Enables seamless integration of multiple function sets, allowing the agent to handle complex, multi-domain queries efficiently. Key features and benefits of Octo-planner: - Efficient on-device planning: Achieves a 97% success rate in in-domain tests while minimizing computational overhead and battery consumption. - Adaptable to new domains: Multi-LoRA merging allows the agent to learn and execute plans across various function sets. - Improved user experience: Enables faster response times and offline functionality. - Enhanced privacy: Keeps data processing local, reducing reliance on cloud services. - Open-source model weights: Encourages further research and development in on-device AI. Octo-planner paves the way for a new generation of efficient and adaptable AI agents that can seamlessly integrate into our daily lives through our personal devices. Read the full paper: https://lnkd.in/gpTW6syn #Octo-planner #PlanningAgents #OnDeviceAI #AIAssistants #Efficiency #Privacy #AI ✔️ Click "Follow" on the Cohorte page for daily AI engineering news. Credits: Wei Chen, Zhiyuan Li, Zhen Guo, Yikang Shen

The Future of Fact-Checking? (Don't Panic Yet!) This mind-blowing video, entirely created with OpenAI's Sora, showcases the incredible power of AI-generated content. It's enough to make you question everything you see online! The future of media consumption might be a wild ride, but it doesn't have to be a descent into chaos. By staying informed and critical of what we see online, we can navigate this new landscape with a healthy dose of skepticism. So, how can you be a responsible digital citizen? - Be source-aware: Question the origin of videos, especially those shared on social media. - Look for inconsistencies: Pay attention to unnatural movements, lighting glitches, or audio-video mismatches in deepfakes. - Use verification tools: Several online tools can help identify potential AI manipulation. Remember, critical thinking is our superpower! By staying vigilant and informed, we can ensure the future of history remains accurate, even in the face of ever-evolving AI technology. _____________ ✔️ Click "Follow" on the Cohorte page for daily AI engineering news. 

🕵️ Can LLMs Help Us Catch LLM Bugs? OpenAI Says Yes! 🐛 OpenAI researchers are tackling a critical challenge in AI development: improving the reliability of human feedback for training large language models (LLMs). Their solution? Train even smarter LLMs to act as "critics" and help humans identify errors in model output. 🤯 Introducing CriticGPT: - Trained with RLHF: CriticGPT is a GPT-4-based model specifically trained to write natural language critiques highlighting problems in code generated by LLMs for real-world assistant tasks. - Outperforms human reviewers: CriticGPT catches more bugs and provides critiques preferred by human annotators over those written by human contractors. - Enhances human evaluation: Human-machine teams combining CriticGPT with human reviewers write more comprehensive critiques while reducing hallucination rates compared to LLMs alone. Key innovations in CriticGPT: - Tampering for improved training data: Contractors adversarially insert subtle bugs into LLM-generated code, creating a dataset of challenging examples with clear reference bugs. - Force Sampling Beam Search (FSBS): Balances the trade-off between comprehensiveness and spurious claims in critiques, enabling fine-grained control over critique quality. - Generalization beyond code: CriticGPT successfully identifies errors in ChatGPT training data for non-code tasks, highlighting its potential for broader applications. This research sheds light on a crucial aspect of scalable oversight – using AI to assist humans in evaluating increasingly complex AI systems. CriticGPT demonstrates the potential of LLM critics to improve the quality and safety of LLM training, paving the way for more robust and trustworthy AI assistants. Read the full paper: https://lnkd.in/dQ8xujfy #CriticGPT #LLM #ScalableOversight #BugDetection #AI #DeepLearning ✔️ Click "Follow" on the Cohorte page for daily AI engineering news. Credits: Nat McAleese, Rai (Michael Pokorny), Juan Felipe Cerón Uribe, Evgenia Nitishinskaya, Maja Trębacz, Jan Leike

The Future of Advertising is Here - McDonald's Goes AI (and You Can Make Ads Too!) Hold onto your ketchup packets, because the future of advertising is here, and it's powered by Artificial Intelligence! Take a peek at this McDonald's commercial created entirely with AI tools. It's a game-changer, and here's why: AI Makes Advertising Easier Than Ever: Most people don't realize how AI can simplify ad creation. This McDonald's commercial is a shining example, built with the following tools: - Midjourney (AI Image Generation): This tool creates stunning visuals from your text descriptions. Imagine generating unique storyboards in seconds! - Magnific AI (Image Enhancement): Say goodbye to grainy footage! Magnific AI upscales and sharpens existing images, making every frame pop. - Luma AI (Dream Machine) - AI Video Generation: This platform is a revolution. Simply feed it text or images, and watch it generate high-quality videos – no cameras or film crews needed! - Runway (AI Video Editing): Real-time editing, motion tracking, and even text-to-video capabilities? Runway streamlines the process, making pro-looking edits a breeze. - ElevenLabs (AI Voice Generation): Need a captivating voiceover in any language? ElevenLabs makes it possible, to add a global touch to your ads. The Benefits of AI Advertising: This isn't just about fancy tech. AI offers a ton of advantages: - No More Specialized Skills: No need for a whole film crew! AI tools make content creation accessible to everyone. - Lightning Speed Production: Gone are the days of lengthy shoots. AI drastically speeds up the creative process. - Cost-Effective Creations: High-quality content without the hefty price tag? Yes, please! AI makes advertising affordable. - Experimentation Playground: Easily try different styles and concepts – the possibilities are endless! - Democratizing the Industry: AI empowers anyone to become a content creator, opening doors for fresh voices. This McDonald's commercial is just a glimpse of what's to come. AI is democratizing advertising, and the potential is limitless! _____________ ✔️ Click "Follow" on the Cohorte page for daily AI engineering news. 

🎉Google Releases Gemma 2: Next-Gen Open Models for AI Development! 🧠 Google has just launched Gemma 2, the latest generation of their open-source, lightweight, and high-performing language models designed for researchers and developers. This release marks a huge step towards making AI more accessible and affordable! Gemma 2 boasts several exciting improvements: - Outsized Performance: The 27B parameter model offers competitive performance to models twice its size, while the 9B model surpasses other open models in its class, including Llama 3 8B. - Unmatched Efficiency: Gemma 2 is optimized for efficient inference at full precision on a single Google Cloud TPU host, NVIDIA A100 or H100 GPU, significantly reducing deployment costs. - Blazing Fast Inference: Experience incredible speed across a range of hardware, from gaming laptops to cloud-based setups, with support for various quantization methods. - Enhanced Developer Experience: Gemma 2 seamlessly integrates with major AI frameworks like Hugging Face Transformers, JAX, PyTorch, and TensorFlow, making it easy to use with your preferred tools. Google emphasizes responsible AI development with Gemma 2: - Commercially-friendly Gemma license: Enables sharing and commercialization of innovations. - Robust safety processes: Pre-training data is carefully filtered, and models undergo rigorous testing for potential biases and risks. - Open-source tooling: The LLM Comparator library and upcoming SynthID watermarking technology provide resources for responsible AI development and deployment. Gemma 2 empowers developers to build ambitious AI projects with cutting-edge performance and efficiency while prioritizing responsible AI practices. Learn more and get started: https://lnkd.in/e-NZTtnT #Gemma2 #GoogleAI #OpenSource #LLMs #AI #DeepLearning ✔️ Click "Follow" on the Cohorte page for daily AI engineering news. Credits: Clement Farabet, Tris Warkentin

Deepfakes are blurring the lines of reality (But How'd They Do That?) Deepfakes are everywhere these days, and they're getting more realistic by the minute! This mind-bending example showcases the power of this technology. But how'd they achieve such a seamless effect? This particular deepfake leverages the impressive combo of ComfyUI and LivePortrait. ComfyUI acts as the mastermind, orchestrating the entire process, while LivePortrait breathes life into the target image by mimicking facial expressions and lip movements in real time. Here's the breakdown: - ComfyUI: This powerful tool acts as the central hub, coordinating the various elements needed for the deepfake. - LivePortrait: The magic happens here! LivePortrait takes a source video and meticulously maps those movements onto a target image, creating the illusion of a living, breathing person. Attribution Matters! It's important to give credit where credit is due. Here's a shoutout to the talented individuals behind this creation: - Image Source: Midjourney – the AI art powerhouse that likely generated the base image. - Source Video: @Juliana Fuentes Silgado – the source of the movements used to animate the deepfake. Credit: @Leo Kadieff – the mastermind who brought it all together. The Future of Deepfakes Deepfakes raise ethical concerns, but their potential for creative applications is undeniable. As technology advances, it'll be crucial to use this power responsibly while appreciating the artistic ingenuity behind these creations. _____________ ✔️ Click "Follow" on the Cohorte page for daily AI engineering news.