Efficient Distributed Training through Gradient Compression with Sparsification and Quantization Techniques