YuliangLiu0306 issues

Results 48 issues of


                                            YuliangLiu0306

[tensor] shape consistency generate transform path and communication cost

In #1418 PR, I add ShapeConsistencyManager into colossalai to support auto parallel strategy search and runtime sharding spec apply. This PR complete the auto parallel strategy search supporting part, mainly...

[tensor] use communication autograd func

Run Build and Test

[autoparallel] add embedding handler

Run Build and Test

[autoparallel] protect bcast handler from invalid strategies

Run Build and Test

[autoparallel] add layernorm handler

Run Build and Test

[autoparallel] add elementwise handler

Run Build and Test

[autoparallel] support distributed dataloader option

Run Build and Test

[autoparallel] remove redundancy comm node

This PR mainly does two things: 1. remove redundancy comm node: if the sharding spec of node output is same as the user node required sharding spec, we will not...

Run Build and Test

[autoparallel] support more flexible data types

### Previous problem The output of majority nodes are a torch.Tensor. However, some kernels, such as split, have a tuple type output, which will lead to the previous implementation broken....

Run Build and Test

[autoparallel] adapt handlers with attention block

### What does this PR do In this PR, fix some bugs on node handlers.

Run Build and Test