Nexa AI

🚀We are thrilled to announce the release of our latest quantized model, Octo Planner GGUF! This cutting-edge model is designed to deliver top-notch performance and efficiency for your AI projects. With reduced memory footprint and faster inference times, it's perfect for both research and production environments. ✨Key Features: - Optimized for speed and efficiency - Suitable for a wide range of AI applications - Seamless integration with existing workflows Check it out here: https://lnkd.in/ga_S5SzB #NexaAI #AI #MachineLearning #QuantizedModel #HuggingFace #AIInnovation #TechNews #AIResearch #ArtificialIntelligence #DeepLearning

🚀 Big News! We're ecstatic to announce that Abubakar Abid, the founder of Gradio, a powerful tool that simplifies building and sharing machine learning models, will be a judge at our upcoming Hackathon hosted by HuggingFace and Nexa AI! 🌟 Showcase your skills, get valuable feedback, and join us for an event full of innovation, creativity, and networking. Don’t miss out! 🔥 Register here➡️: https://lnkd.in/gEkprgkm #Hackathon #AI #MachineLearning #Gradio #NexaAI #Innovation #TechEvent #huggingface

Join the Super AI Agent Hackathon hosted by Nexa AI and Hugging Face! Compete to win up to $7,000 in cash prizes by building the best AI-powered app. 1st Round: July 19 - August 18 | Online 2nd Round: August 25, 2024 | The Hub @ Stanford Research Park Contact: octopus@nexa4ai.com More info: https://lnkd.in/gUrv84sa

Check out how our Coral SDK performs on the Samsung S23! Watch the video to see the impressive speed and token generation capabilities. #AI #SDK #EdgeComputing #MachineLearning #AIDevelopment #CoralSDK Samsung Electronics America Samsung Electronics Samsung Developer

Check out how our Coral SDK performs on the Samsung S23! Watch the video to see the impressive speed and token generation capabilities. #AI #SDK #EdgeComputing #MachineLearning #AIDevelopment #CoralSDK Samsung Electronics America Samsung Electronics Samsung Developer

Serving large language models (LLMs) directly on devices has long been a challenge in the industry due to memory, battery, and computational restrictions. Nexa AI's Coral SDK achieves 5X faster time-to-first-token and 20% decoding speed up than Llama-cpp and MLC-LLM for inference on CPU. See our SDK at https://www.nexa4ai.com/

Our Coral SDK achieves impressive inference speeds, faster than Llama-cpp and MLC-LLM for inference on CPU, and we run the benchmark on Phi3 on Google Pixel 6. Coral SDK supports Android, iOS, MacOS, and Microsoft Windows with multi-processor and flexible quantization options. Discover seamless integration of on-device LLMs into your devices! 👉 Learn more: https://lnkd.in/gR6mupFF #AI #SDK #EdgeComputing #CoralSDK #llamacpp #MLC #Phi3 #Google #Pixel6

🚀 Introducing Octo-planner: a 3B parameter, on-device AI planning model by Nexa AI, achieving high accuracy and efficiency with a Planner-Action framework, in collaboration with Massachusetts Institute of Technology and IBM Watson AI Lab. 🚀 Octo-planner Highlights: 🌟 On-device planning: Octo-planner, a 3.8B parameter model optimized for edge devices. 🛠️ Planner-Action framework: Distinct components for planning and action execution. 📈 Highly accurate: 97% success rate in in-domain tests. 🌐 Multi-domain capability: Novel multi-LoRA training method for complex queries. Paper：https://lnkd.in/gpTW6syn Demo：https://lnkd.in/gWpaugqD Model Page: https://lnkd.in/gmRYJyWG #AI #EdgeComputing #Innovation #TechNews

We are excited to announce that Nexa AI has joined the NVIDIA Inception program! This recognition supports our mission to transform industries with our innovative Octopus AI models, developed by our founders Wei Chen Alex (Wei) Chen and Zack Li Zack Li. Thanks to NVIDIA for the support! #NVIDIAInception #AI #Innovation #NexaAI

🚀 This summer, we Nexa AI with Huggingface will host an AI Agent Hackathon at Stanford Research Park ✨ Let's build something remarkable together! #AI #Hackathon #Innovation #NexaAIxHuggingface