Ternary Neural Networks with Fine-Grained Quantization | Read Paper on Bytez