v0.5.0
Closed Jan 29, 2022
100% complete
The first anniversary of open source of OneFlow.
Eager new features:
- Support Eager Mirror paralllel strategy in multi-devices
- Support Eager Consistent paralllel strategy in multi-devices
- nn.Module add a lot of ops
Fully compatible with pytorch interface, just import oneflow as torch
.
- flow.experimental.xx -> flow.xx
- Old versions of interfaces (like tens…
The first anniversary of open source of OneFlow.
Eager new features:
- Support Eager Mirror paralllel strategy in multi-devices
- Support Eager Consistent paralllel strategy in multi-devices
- nn.Module add a lot of ops
Fully compatible with pytorch interface, just import oneflow as torch
.
- flow.experimental.xx -> flow.xx
- Old versions of interfaces (like tensorflow 1.x) will no longer be supported
- Run OneFlow Lazy Consistent distribute training with 1.0 interface API (flow.nn.Graph)
This milestone is closed.
No open issues remain. View closed issues or see open milestones in this repository.