为Spark Deep Learning 添加NLP处理实现_51CTO博客_3rules to...
现有的分布式调参功能,基本不可用。参看我提的这个Issue: To Avoid collecting trainning data to driver and broadcasting them 不支持分布式tranning. 参看我提的这个Issue: Is there any plan to port TensorframeOnSpark(From yahoo) ? 当然SDL的想法非常好: 相比K8s + TF只是完成...