alibaba/EasyParallelLibrary

Issues

• Has there been an improvement in single-GPU training speed?

  #31 opened 3 months ago by a1342772
  1

• 2台服务器分布式跑example中的resnet_split.py遇到无限等待的情况

  #28 opened 8 months ago by alphabewitch
  4

• epl单机单卡和单机多卡训练step如何理解

  #30 opened a year ago by SueeH
  1

• 2机2卡实验NCCL报错

  #29 opened a year ago by wind818
  1

• AttributeError: 'NoneType' object has no attribute 'taskgraph'

  #27 opened 2 years ago by Waterpine
  3

• Problem of Data Parallel Model, program didn't end when reached global step

  #20 opened 2 years ago by Jimmy-jin
  1

• DingTalk QR code is outdated

  #25 opened 2 years ago by co63oc
  2

• DistributedDense只支持按照列切分吗？

  #22 opened 2 years ago by kuangdao
  1

• Gradient Checkpoint with auto type got a TypeError

  #19 opened 2 years ago by RussellZZ
  1

• 训练时，除chief worker外，其余worker在每次save checkpoint 后 step归0，且在第二次save checkpoint 后 整个进程卡死

  #17 opened 2 years ago by walkingwindy
  1