24 Commits (c257cf36e2ff8b2c19ccdd39ca6aff6f66cd7e3b)

Author SHA1 Message Date
  yao_yf dc7dc7d3fa dataset strategy set 4 years ago
  Xiaoda Zhang bb5d4212f7 enable All2All in infering redistribution ops 4 years ago
  Xiaoda Zhang 04381273b3 Add the sharding propagation function: 5 years ago
  yao_yf d2dc22ff71 parallel_weight_init_adapt_pipeline_increment_predict 4 years ago
  i-robot b79400296a !17959 fix optimizer weight shard config 4 years ago
  kswang 8aa0450b8d set dtype for allreduce fusion 4 years ago
  Ziyan 95ac0f6d58 fix optimizer weight shard config 4 years ago
  Ziyan 2a752f24bf enable not fully use opt shard 5 years ago
  liujunzhu 6541b96c40 Add communication parallel mode. 5 years ago
  yangzhenzhang a70d616841 mini step grad accumulation 5 years ago
  yangzhenzhang 7303c3d3b8 add group ckpt 5 years ago
  yangzhenzhang 9da3f9bec9 mini step grad accumulation 5 years ago
  yangzhenzhang 278e82a849 update pipeline parallel 5 years ago
  lichenever 2e1c43483e add auto parallel pipeline 5 years ago
  Yi Huaijie d7faa77b5e support int64 shape 5 years ago
  Xiaoda Zhang fba2bfeb54 overwrite strategies for star graph structure 5 years ago
  huangxinjing 4ef439e27b Add stage information for ops and strategy 5 years ago
  yao_yf d4cfe55c04 rename mirror_mean to gradients_mean 5 years ago
  yao_yf 8f7aa5bd5a auto parallel context modify 5 years ago
  Yi Huaijie 394be43492 raise RuntimeError when set different mode after Initializer created 5 years ago
  Yi Huaijie 89a4ebf8a1 parallel mode must be set before create an initializer 5 years ago
  liubuyu a499d4e47c decoupling core and debug 5 years ago
  liubuyu 76dc80e7b7 Unified code style 5 years ago
  liubuyu 43c79eb853 mindspore path adjust 5 years ago