[dynamo][pytree][2/N] make CXX pytree traceable: tree_flatten / tree_unflatten / tree_structure
#137398
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
This was referenced Oct 5, 2024
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/137398
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 17d9823 with merge base 7435f57 ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
This was referenced Oct 5, 2024
XuehaiPan
added a commit
to XuehaiPan/pytorch
that referenced
this pull request
Oct 5, 2024
…ee_unflatten` / `tree_structure` ghstack-source-id: 7deed05 Pull Request resolved: pytorch#137398
XuehaiPan
added a commit
to XuehaiPan/pytorch
that referenced
this pull request
Oct 5, 2024
…ee_unflatten` / `tree_structure` ghstack-source-id: ad8c86d Pull Request resolved: pytorch#137398
XuehaiPan
added a commit
to XuehaiPan/pytorch
that referenced
this pull request
Oct 5, 2024
…ee_unflatten` / `tree_structure` ghstack-source-id: a47eb32 Pull Request resolved: pytorch#137398
XuehaiPan
added a commit
to XuehaiPan/pytorch
that referenced
this pull request
Oct 5, 2024
…ee_unflatten` / `tree_structure` ghstack-source-id: d441e86 Pull Request resolved: pytorch#137398
XuehaiPan
added a commit
to XuehaiPan/pytorch
that referenced
this pull request
Oct 5, 2024
…ee_unflatten` / `tree_structure` ghstack-source-id: 1025e14 Pull Request resolved: pytorch#137398
The existing C++ pytree and Python pytree share the same API and function signature. There are tests in our CI to ensure this. More, the pytree tests and dynamo tests are both tested against the Python pytree, C++ pytree, and the new public pytree API (a thin wrapper around the underlying pytree implementation). |
|
Hi @jansel, could you take a look at this PR and the next PR in the stack? Then the follow-ups can be reviewed and merged separately. |
jansel
approved these changes
Dec 12, 2024
pytorchmergebot
pushed a commit
that referenced
this pull request
Dec 12, 2024
…ap_` (#137399) Pull Request resolved: #137399 Approved by: https://github.com/jansel ghstack dependencies: #137398
bluenote10
pushed a commit
to bluenote10/pytorch
that referenced
this pull request
Dec 14, 2024
…ee_unflatten` / `tree_structure` (pytorch#137398) Pull Request resolved: pytorch#137398 Approved by: https://github.com/jansel
bluenote10
pushed a commit
to bluenote10/pytorch
that referenced
this pull request
Dec 14, 2024
…ap_` (pytorch#137399) Pull Request resolved: pytorch#137399 Approved by: https://github.com/jansel ghstack dependencies: pytorch#137398
XuehaiPan
added a commit
to XuehaiPan/pytorch
that referenced
this pull request
Dec 18, 2024
…ee_unflatten` / `tree_structure` (pytorch#137398) Pull Request resolved: pytorch#137398 Approved by: https://github.com/jansel
XuehaiPan
added a commit
to XuehaiPan/pytorch
that referenced
this pull request
Dec 18, 2024
…ap_` (pytorch#137399) Pull Request resolved: pytorch#137399 Approved by: https://github.com/jansel ghstack dependencies: pytorch#137398
This was referenced Dec 18, 2024
XuehaiPan
added a commit
to XuehaiPan/pytorch
that referenced
this pull request
Dec 19, 2024
…ee_unflatten` / `tree_structure` (pytorch#137398) Pull Request resolved: pytorch#137398 Approved by: https://github.com/jansel
XuehaiPan
added a commit
to XuehaiPan/pytorch
that referenced
this pull request
Dec 19, 2024
…ap_` (pytorch#137399) Pull Request resolved: pytorch#137399 Approved by: https://github.com/jansel ghstack dependencies: pytorch#137398
XuehaiPan
added a commit
to XuehaiPan/pytorch
that referenced
this pull request
Dec 24, 2024
…ee_unflatten` / `tree_structure` (pytorch#137398) Pull Request resolved: pytorch#137398 Approved by: https://github.com/jansel
XuehaiPan
added a commit
to XuehaiPan/pytorch
that referenced
this pull request
Dec 24, 2024
…ap_` (pytorch#137399) Pull Request resolved: pytorch#137399 Approved by: https://github.com/jansel ghstack dependencies: pytorch#137398
XuehaiPan
added a commit
to XuehaiPan/pytorch
that referenced
this pull request
Dec 25, 2024
…ee_unflatten` / `tree_structure` (pytorch#137398) Pull Request resolved: pytorch#137398 Approved by: https://github.com/jansel
XuehaiPan
added a commit
to XuehaiPan/pytorch
that referenced
this pull request
Dec 25, 2024
…ap_` (pytorch#137399) Pull Request resolved: pytorch#137399 Approved by: https://github.com/jansel ghstack dependencies: pytorch#137398
XuehaiPan
added a commit
to XuehaiPan/pytorch
that referenced
this pull request
Dec 30, 2024
…ee_unflatten` / `tree_structure` (pytorch#137398) Pull Request resolved: pytorch#137398 Approved by: https://github.com/jansel
XuehaiPan
added a commit
to XuehaiPan/pytorch
that referenced
this pull request
Dec 30, 2024
…ap_` (pytorch#137399) Pull Request resolved: pytorch#137399 Approved by: https://github.com/jansel ghstack dependencies: pytorch#137398
XuehaiPan
added a commit
to XuehaiPan/pytorch
that referenced
this pull request
Jan 7, 2025
…ee_unflatten` / `tree_structure` (pytorch#137398) Pull Request resolved: pytorch#137398 Approved by: https://github.com/jansel
XuehaiPan
added a commit
to XuehaiPan/pytorch
that referenced
this pull request
Jan 7, 2025
…ap_` (pytorch#137399) Pull Request resolved: pytorch#137399 Approved by: https://github.com/jansel ghstack dependencies: pytorch#137398
XuehaiPan
added a commit
to XuehaiPan/pytorch
that referenced
this pull request
Jan 10, 2025
…ee_unflatten` / `tree_structure` (pytorch#137398) Pull Request resolved: pytorch#137398 Approved by: https://github.com/jansel
XuehaiPan
added a commit
to XuehaiPan/pytorch
that referenced
this pull request
Jan 10, 2025
…ap_` (pytorch#137399) Pull Request resolved: pytorch#137399 Approved by: https://github.com/jansel ghstack dependencies: pytorch#137398
XuehaiPan
added a commit
to XuehaiPan/pytorch
that referenced
this pull request
Jan 15, 2025
…ee_unflatten` / `tree_structure` (pytorch#137398) Pull Request resolved: pytorch#137398 Approved by: https://github.com/jansel
XuehaiPan
added a commit
to XuehaiPan/pytorch
that referenced
this pull request
Jan 15, 2025
…ap_` (pytorch#137399) Pull Request resolved: pytorch#137399 Approved by: https://github.com/jansel ghstack dependencies: pytorch#137398
zou3519
reviewed
Feb 11, 2025
Comment on lines
+115
to
+116
| @dataclass(frozen=True) | F438||
| class PyTreeSpec: |
There was a problem hiding this comment.
@XuehaiPan, if I'm understanding this correctly, you're adding a polyfill for cxx_pytree.tree_flatten such that it will return an instance of this PyTreeSpec class. I'm not sure this works if we are trying to return a PyTreeSpec from a torch.compile'd function: does it create an instance of this "class PyTreeSpec" object, or does it create an instance of torch.utils._cxx_pytree.TreeSpec?
There was a problem hiding this comment.
In a compiled function, it returns torch._dynamo.polyfills.pytree.PyTreeSpec. This class provides the exactly same interfaces with torch.utils._cxx_pytree.TreeSpec.
There was a problem hiding this comment.
Yeah this is wrong, see the following:
import torch
import torch.utils._cxx_pytree as pytree
@torch.compile(backend="eager", fullgraph=True)
def f(x, y):
vals, spec = pytree.tree_flatten(x)
return vals, spec, y.sin()
y = torch.randn(3)
x = [1, [2, [3, 4]]]
vals, spec, _ = f(x, y)
this_doesnt_work = pytree.tree_unflatten(vals, spec)There was a problem hiding this comment.
I see. It may cause problems if we only compile part of the program.
There is an in-progress polyfill infra for C++ classes.
There was a problem hiding this comment.
Opened a PR to fix this:
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):
torch.utils.pytreeby default #138056torch.func#137884torch.utils.pytree#137400dictkeys in insertion order in CXX pytree #130140tree_map/tree_map_#137399tree_flatten/tree_unflatten/tree_structure#137398cc @zou3519 @voznesenskym @penguinwu @EikanWang @jgong5 @Guobing-Chen @XiaobingSuper @zhuhaozhe @blzheng @wenzhe-nrv @jiayisunx @chenyang78 @kadeng @chauhang @amjames @rec