Founded in 2016,Deephi Tech aims to become the recognized leader in deep learning accleration. We provide an end-to-end solution utilizing deep compression and a DPU platform.Leveraging the optimization of co-designed neural networks and FPGAs,Deephi provides more efficient,convenient and economical inference platforms for both edges and sever-side,in industries including but not limited to data centers and surveillance.
This is a model quantization tool for convolution neural networks(CNN) in caffemodel format, which converts parameters from FP32 format to INT8 format or other bit depths. This tool supports common layer types in neural networks and is optimized for DeePhi DPU.
Linux/Unix, Ubuntu 4.4.0-1062-aws - 64-bit Amazon Machine Image (AMI)
showing 1 - 1