The training module of SMT involves a large amount of corpus which was usually implemented in mainframe servers.

• 统计机器翻译模型的训练需要处理大量训练语料，这通常需要大型服务器才能实现。