[FSDP][5/N] Unblock ignored_states + auto wrap (for now)
#104418
Closed
Conversation
This file contains 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
[ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/104418
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 8b25684: This comment was automatically generated by Dr. CI and updates every 15 minutes. |
pytorch-bot
bot
added
the
release notes: distributed (fsdp)
This was referenced Jun 29, 2023
awgu
added a commit
that referenced
this pull request
Jun 29, 2023
ghstack-source-id: 38b3b336679415d39fcafd4c8c4c4d707e3816c5 Pull Request resolved: #104418
The "for now" is because we still have the issue that when using the parameter `ignored_states` path, we do not recover the ignored modules, so FSDP still wraps those as empty shells (no managed parameters), which is not ideal. This is not a blocking issue as far as I know. [ghstack-poisoned]
awgu
added a commit
that referenced
this pull request
Jun 29, 2023
ghstack-source-id: 3dece821cc9b5b6fac09519f56414fbcf4ca3874 Pull Request resolved: #104418
The "for now" is because we still have the issue that when using the parameter `ignored_states` path, we do not recover the ignored modules, so FSDP still wraps those as empty shells (no managed parameters), which is not ideal. This is not a blocking issue as far as I know. [ghstack-poisoned]
awgu
added a commit
that referenced
this pull request
Jun 29, 2023
ghstack-source-id: 7a743d525ebfeee9362719c13dcef95993ecb0fa Pull Request resolved: #104418
awgu
added a commit
to awgu/pytorch
that referenced
this pull request
Jun 29, 2023
ghstack-source-id: 7a743d525ebfeee9362719c13dcef95993ecb0fa Pull Request resolved: pytorch#104418
The "for now" is because we still have the issue that when using the parameter `ignored_states` path, we do not recover the ignored modules, so FSDP still wraps those as empty shells (no managed parameters), which is not ideal. This is not a blocking issue as far as I know. [ghstack-poisoned]
awgu
added a commit
to awgu/pytorch
that referenced
this pull request
Jun 30, 2023
ghstack-source-id: 60be363bf4780ea201280664d0fe0ede7e7faaa5 Pull Request resolved: pytorch#104418
The "for now" is because we still have the issue that when using the parameter `ignored_states` path, we do not recover the ignored modules, so FSDP still wraps those as empty shells (no managed parameters), which is not ideal. This is not a blocking issue as far as I know. [ghstack-poisoned]
awgu
added a commit
to awgu/pytorch
that referenced
this pull request
Jun 30, 2023
ghstack-source-id: 28c95ac2e70c519db0e32cd6b3147ede9aed54cd Pull Request resolved: pytorch#104418
The "for now" is because we still have the issue that when using the parameter `ignored_states` path, we do not recover the ignored modules, so FSDP still wraps those as empty shells (no managed parameters), which is not ideal. This is not a blocking issue as far as I know. [ghstack-poisoned]
The "for now" is because we still have the issue that when using the parameter `ignored_states` path, we do not recover the ignored modules, so FSDP still wraps those as empty shells (no managed parameters), which is not ideal. This is not a blocking issue as far as I know. [ghstack-poisoned]
awgu
added a commit
to awgu/pytorch
that referenced
this pull request
Jun 30, 2023
ghstack-source-id: eee579391a4b12b8558a922a2a17cf93f9289b90 Pull Request resolved: pytorch#104418
The "for now" is because we still have the issue that when using the parameter `ignored_states` path, we do not recover the ignored modules, so FSDP still wraps those as empty shells (no managed parameters), which is not ideal. This is not a blocking issue as far as I know. [ghstack-poisoned]
awgu
added a commit
to awgu/pytorch
that referenced
this pull request
Jul 5, 2023
ghstack-source-id: 5d23b076f1edb0ecb0b441f14fb7732e2be2d7e1 Pull Request resolved: pytorch#104418
awgu
added a commit
to awgu/pytorch
that referenced
this pull request
Jul 5, 2023
ghstack-source-id: 5d23b076f1edb0ecb0b441f14fb7732e2be2d7e1 Pull Request resolved: pytorch#104418
The "for now" is because we still have the issue that when using the parameter `ignored_states` path, we do not recover the ignored modules, so FSDP still wraps those as empty shells (no managed parameters), which is not ideal. This is not a blocking issue as far as I know. [ghstack-poisoned]
awgu
requested review from
zhaojuanmao,
rohan-varma,
H-Huang,
kwen2501,
wanchaol,
fegin,
fduwjj,
kiukchung and
d4l3k
as code owners
The "for now" is because we still have the issue that when using the parameter `ignored_states` path, we do not recover the ignored modules, so FSDP still wraps those as empty shells (no managed parameters), which is not ideal. This is not a blocking issue as far as I know. [ghstack-poisoned]
This was referenced Jul 6, 2023
The "for now" is because we still have the issue that when using the parameter `ignored_states` path, we do not recover the ignored modules, so FSDP still wraps those as empty shells (no managed parameters), which is not ideal. This is not a blocking issue as far as I know. [ghstack-poisoned]
rohan-varma
reviewed
Jul 6, 2023
| @@ -417,6 +417,7 @@ def __init__( | |||
| "forward_prefetch": forward_prefetch, | |||
| "limit_all_gathers": limit_all_gathers, | |||
| "use_orig_params": use_orig_params, | |||
| "ignored_states": self._ignored_params, | |||
There was a problem hiding this comment.
why do we need to propagate ignored_states, but not ignored_modules?
There was a problem hiding this comment.
Good point. Maybe we should also propagate ignored_modules. This might be another bug.
There was a problem hiding this comment.
The reason not propagating ignored_modules does not break correctness today is because we only use ignored_modules to compute the ignored parameters and to compute which modules to ignore for auto wrapping.
For those two functionalities, we do not need the nested FSDP instances to have ignored_modules. However, we should probably still propagate it in case we use ignored_modules for something else in the future.
There was a problem hiding this comment.
I see. In general, what do you think about having both ignored_states and ignored_modules? this seems like it can get confusing to users, shall we just consolidate to ignored_states?
| @@ -173,29 +165,32 @@ def _test_ignored_modules_transformer( | |||
| CUDAInitMode.CUDA_BEFORE, | |||
| deterministic=True, | |||
| ) | |||
| if use_auto_wrap: | |||
| nonwrapped_model.output_proj.weight = nn.Parameter( | |||
There was a problem hiding this comment.
why do we need this reassignment?
There was a problem hiding this comment.
See the comment above about unsharing the weight.
There was a problem hiding this comment.
I can duplicate the comment here if it helps.
The "for now" is because we still have the issue that when using the parameter `ignored_states` path, we do not recover the ignored modules, so FSDP still wraps those as empty shells (no managed parameters), which is not ideal. This is not a blocking issue as far as I know. [ghstack-poisoned]
The "for now" is because we still have the issue that when using the parameter `ignored_states` path, we do not recover the ignored modules, so FSDP still wraps those as empty shells (no managed parameters), which is not ideal. This is not a blocking issue as far as I know. [ghstack-poisoned]
rohan-varma
approved these changes
Jul 7, 2023
The "for now" is because we still have the issue that when using the parameter `ignored_states` path, we do not recover the ignored modules, so FSDP still wraps those as empty shells (no managed parameters), which is not ideal. This is not a blocking issue as far as I know. [ghstack-poisoned]
|
@pytorchbot merge |
Merge startedYour change will be merged once all checks pass (ETA 0-4 Hours). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
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.
Stack from ghstack (oldest at bottom):
ModuleWrapPolicy#104427ignored_states+ auto wrap (for now) #104418_get_fully_sharded_module_to_states#104409fully_shardauto wrap #104408_auto_wrapforfully_shard#104407ModuleWrapPolicyto new path #104346The "for now" is because we still have the issue that when using the parameter
ignored_statespath, we do not recover the ignored modules, so FSDP still wraps those as empty shells (no managed parameters), which is not ideal. This is not a blocking issue as far as I know.