The MERA Compiler and Framework by EdgeCortix is a groundbreaking software platform crafted to facilitate AI inference across heterogeneous computing systems. It enables seamless integration and deployment of AI models using the Dynamic Neural Accelerator and supports complex generative AI applications for vision, audio, and language tasks.
Offering a comprehensive suite of tools and APIs, MERA aids in graph compilation, model calibration, and quantization, ensuring the rapid deployment of pre-trained neural networks. The framework supports multiple machine learning frameworks, such as PyTorch, TensorFlow Lite, and ONNX, empowering developers to efficiently adapt AI solutions without delving into intricate hardware specifics.
MERA stands out with its inherent support for diverse processors, such as AMD, Intel, and Arm, providing flexibility and compatibility across various hardware environments. This platform enhances development speed and reduces time-to-market, delivering high-performance AI inference solutions that are energy-efficient and versatile for edge AI applications.
