Fork自 PaddlePaddle / Paddle
* Fix average_accumulate_op for parallel executor. * Fix model average on multi-GPUs.
拖放文件到此处或点击上传