[hybrid bug fix] Fix mp multi gradient clip prob #35713
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
Thanks for your contribution! |
wangxicoding
requested changes
Sep 14, 2021
python/paddle/distributed/fleet/meta_optimizers/sharding/gradient_clip_helper.py
Show resolved
Hide resolved
wangxicoding
requested changes
Sep 14, 2021
python/paddle/distributed/fleet/meta_optimizers/sharding/gradient_clip_helper.py
Outdated
Show resolved
Hide resolved
python/paddle/distributed/fleet/meta_optimizers/sharding/gradient_clip_helper.py
Outdated
Show resolved
Hide resolved
python/paddle/distributed/fleet/meta_optimizers/sharding/gradient_clip_helper.py
Show resolved
Hide resolved
python/paddle/distributed/fleet/meta_optimizers/sharding/gradient_clip_helper.py
Outdated
Show resolved
Hide resolved
5e1fe33 to
7ebecff
Compare
|
|
python/paddle/distributed/fleet/meta_optimizers/sharding/gradient_clip_helper.py
Show resolved
Hide resolved
| # Therefore, we prune those duplicated vars for grad clip. | ||
| if mp_rank > 0 and (not (hasattr(input_var, 'is_distributed') | ||
| and input_var.is_distributed)): | ||
| removed_op_idx.append(idx) |
There was a problem hiding this comment.
其实当前这种删法是不严谨的，按照之前global_clip的square、sum op实现会出问题。不过目前改成了squarel2norm实现，不会出问题就是了。
python/paddle/distributed/fleet/meta_optimizers/sharding/gradient_clip_helper.py
Outdated
Show resolved
Hide resolved
python/paddle/distributed/fleet/meta_optimizers/sharding/gradient_clip_helper.py
Outdated
Show resolved
Hide resolved
python/paddle/distributed/fleet/meta_optimizers/sharding/gradient_clip_helper.py
Show resolved
Hide resolved
a9763af to
ee1eff2
Compare
python/paddle/distributed/fleet/meta_optimizers/sharding/gradient_clip_helper.py
Outdated
Show resolved
Hide resolved
70baf73 to
d9b6d50
Compare
wangxicoding
approved these changes
Sep 16, 2021
| 'cast', 'sum', 'fill_constant', 'cast', 'sum', 'fill_constant', | ||
| 'cast', 'sum', 'c_sync_comm_stream', 'check_finite_and_unscale', | ||
| 'cast', 'c_allreduce_max', 'c_allreduce_max', 'cast', | ||
| 'update_loss_scaling', 'fill_constant', 'c_allreduce_sum', |
| OP_ROLE_KEY: OpRole.Optimize, | ||
| }) | ||
| return | ||
| for idx, op in list(enumerate(block.ops)): |
AnnaTrainingG
pushed a commit
to AnnaTrainingG/Paddle
that referenced
this pull request
Sep 29, 2021
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
PR types
Bug fixes
PR changes
Others
Describe
mp下，有些var并不是分布式的，譬如scale、bias等等。这些var如果在GradientClipByGlobalNorm时候在各路mp均进行累加，那么最后的global norm会略大一些。
修改后，对于is_distributed为False的grad，只在mp_rank为0的节点进行累加计算Norm，后续通过c_allreduce_sum来计算GlobalNorm
修改前mp_rank=1 program:
修改后mp_rank=1 program:
