Update README.md

README.md

@@ -6,10 +6,10 @@ colorTo: green
 sdk: static
 pinned: true
 ---
-**GreenBitAI** is dedicated to advancing cutting-edge technologies in the field of artificial intelligence and we advocate for sustainable machine learning practices, ensuring that AI technologies have a positive impact on our environment and society. 
-Our primary focus areas include:
-- Low-Bit Neural Networks: We specialize in developing and optimizing low-bit neural networks, making AI models more efficient and accessible.
-- Edge AI: We are passionate about pushing the boundaries of AI to run efficiently on edge devices, enabling intelligent applications in resource-constrained environments.
+**GreenBitAI** is at the forefront of advancing cutting-edge technologies in the field of artificial intelligence, championing sustainable machine learning practices to ensure a positive impact of AI technologies on our environment and society. They contribute to the open-source community by continually offering the latest models and advanced software tools. 
+Their primary focus areas include:
 
-We are committed to contributing to the open-source community by continually providing state-of-the-art models and cutting-edge software tools.
-Join us in our mission to make AI more sustainable, efficient, and widely accessible.
+- Low-Bit Neural Networks: Specializing in the development and optimization of low-bit neural networks, GreenBitAI makes AI models more efficient and accessible.
+- Edge AI and Cloud-Edge Collaboration: They are dedicated to efficiently running AI on edge devices while exploring the integration of cloud and edge computing. This approach aims to achieve efficient collaboration between cloud and edge, combining the computational power of the cloud with the real-time response advantages of edge devices. This integrated method provides flexible and efficient solutions for various applications.
+
+Recently, GreenBitAI has successfully released 4-bit and 2-bit models ranging from TinyLLaMA 1.1B to LLaMA 70B. Despite extreme compression, these models maintain robust performance and are available in the HuggingFace model repository. Notably, they have also open-sourced 4-bit versions of the 01-Yi 34B and Yi 6B models. After testing by the 01-Yi team on 21 mainstream LLM tasks, GreenBitAI's 4-bit versions show less than a 1% performance gap compared to the official 16-bit versions, with detailed assessment results available on the model page. Moreover, GreenBitAI has announced the upcoming release of new 2-bit models with similarly near-lossless efficiency, which is highly anticipated.