High-performance Tensor Processing Unit (TPU) implementation on FPGA with multi-precision support (INT4/INT8/FP16/FP32) - View it on GitHub