[FX] Add Interpreter and Transformer #50420
Closed
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
[ghstack-poisoned]
This was referenced Jan 12, 2021
💊 CI failures summary and remediationsAs of commit 64d7a19 (more details on the Dr. CI page):
This comment was automatically generated by Dr. CI (expand for details).Follow this link to opt-out of these comments for your Pull Requests.Please report bugs/suggestions to the (internal) Dr. CI Users group. |
[ghstack-poisoned]
|
I think it might be easier to implement vmap with this. |
1. Codify the interpreter pattern that was used, for ex, in [shape prop](https://github.com/pytorch/pytorch/blob/c3b4b2062748ed70e689f9f9ffe670e6fa20a071/torch/fx/experimental/shape_prop.py#L7). This is a pattern users will likely find repeatedly useful so it's good to have it first-class in the API. As seen in the test cases, this makes it incredibly easy to do simple analysis and transforms like swapping out individual nodes 2. Add a small `Transformer` class that is an `Interpreter` but provides a `transform` method that yields a GraphModule from interpreting the code TODO: - [ ] Write docstrings - [ ] Write docs - [ ] Look for more places this can be used Differential Revision: [D25880330](https://our.internmc.facebook.com/intern/diff/D25880330) [ghstack-poisoned]
1. Codify the interpreter pattern that was used, for ex, in [shape prop](https://github.com/pytorch/pytorch/blob/c3b4b2062748ed70e689f9f9ffe670e6fa20a071/torch/fx/experimental/shape_prop.py#L7). This is a pattern users will likely find repeatedly useful so it's good to have it first-class in the API. As seen in the test cases, this makes it incredibly easy to do simple analysis and transforms like swapping out individual nodes 2. Add a small `Transformer` class that is an `Interpreter` but provides a `transform` method that yields a GraphModule from interpreting the code TODO: - [ ] Write docstrings - [ ] Write docs - [ ] Look for more places this can be used Differential Revision: [D25880330](https://our.internmc.facebook.com/intern/diff/D25880330) [ghstack-poisoned]
1. Codify the interpreter pattern that was used, for ex, in [shape prop](https://github.com/pytorch/pytorch/blob/c3b4b2062748ed70e689f9f9ffe670e6fa20a071/torch/fx/experimental/shape_prop.py#L7). This is a pattern users will likely find repeatedly useful so it's good to have it first-class in the API. As seen in the test cases, this makes it incredibly easy to do simple analysis and transforms like swapping out individual nodes 2. Add a small `Transformer` class that is an `Interpreter` but provides a `transform` method that yields a GraphModule from interpreting the code TODO: - [ ] Write docstrings - [ ] Write docs - [ ] Look for more places this can be used Differential Revision: [D25880330](https://our.internmc.facebook.com/intern/diff/D25880330) [ghstack-poisoned]
1. Codify the interpreter pattern that was used, for ex, in [shape prop](https://github.com/pytorch/pytorch/blob/c3b4b2062748ed70e689f9f9ffe670e6fa20a071/torch/fx/experimental/shape_prop.py#L7). This is a pattern users will likely find repeatedly useful so it's good to have it first-class in the API. As seen in the test cases, this makes it incredibly easy to do simple analysis and transforms like swapping out individual nodes 2. Add a small `Transformer` class that is an `Interpreter` but provides a `transform` method that yields a GraphModule from interpreting the code TODO: - [x] Write docstrings - [ ] Write docs - [ ] Look for more places this can be used Differential Revision: [D25880330](https://our.internmc.facebook.com/intern/diff/D25880330) [ghstack-poisoned]
1. Codify the interpreter pattern that was used, for ex, in [shape prop](https://github.com/pytorch/pytorch/blob/c3b4b2062748ed70e689f9f9ffe670e6fa20a071/torch/fx/experimental/shape_prop.py#L7). This is a pattern users will likely find repeatedly useful so it's good to have it first-class in the API. As seen in the test cases, this makes it incredibly easy to do simple analysis and transforms like swapping out individual nodes 2. Add a small `Transformer` class that is an `Interpreter` but provides a `transform` method that yields a GraphModule from interpreting the code   TODO: - [x] Write docstrings - [ ] Write docs - [ ] Look for more places this can be used Differential Revision: [D25880330](https://our.internmc.facebook.com/intern/diff/D25880330) [ghstack-poisoned]
1. Codify the interpreter pattern that was used, for ex, in [shape prop](https://github.com/pytorch/pytorch/blob/c3b4b2062748ed70e689f9f9ffe670e6fa20a071/torch/fx/experimental/shape_prop.py#L7). This is a pattern users will likely find repeatedly useful so it's good to have it first-class in the API. As seen in the test cases, this makes it incredibly easy to do simple analysis and transforms like swapping out individual nodes 2. Add a small `Transformer` class that is an `Interpreter` but provides a `transform` method that yields a GraphModule from interpreting the code   TODO: - [x] Write docstrings - [ ] Write docs - [ ] Look for more places this can be used Differential Revision: [D25880330](https://our.internmc.facebook.com/intern/diff/D25880330) [ghstack-poisoned]
zdevito
reviewed
Jan 29, 2021
| .. autoclass:: torch.fx.Interpreter | ||
| :members: | ||
|
|
||
| .. autoclass:: torch.fx.Transformer |
There was a problem hiding this comment.
still need a different name for this
There was a problem hiding this comment.
I haven't heard any good suggestions for this. Is the concern that this is the same name as the Transformer architecture? In that case, I would argue that "Transformer" is a terrible name for that architecture and I don't think we should let them squat on an English language common noun
1. Codify the interpreter pattern that was used, for ex, in [shape prop](https://github.com/pytorch/pytorch/blob/c3b4b2062748ed70e689f9f9ffe670e6fa20a071/torch/fx/experimental/shape_prop.py#L7). This is a pattern users will likely find repeatedly useful so it's good to have it first-class in the API. As seen in the test cases, this makes it incredibly easy to do simple analysis and transforms like swapping out individual nodes 2. Add a small `Transformer` class that is an `Interpreter` but provides a `transform` method that yields a GraphModule from interpreting the code   TODO: - [x] Write docstrings - [x] Write docs - [ ] Look for more places this can be used Differential Revision: [D25880330](https://our.internmc.facebook.com/intern/diff/D25880330) [ghstack-poisoned]
1. Codify the interpreter pattern that was used, for ex, in [shape prop](https://github.com/pytorch/pytorch/blob/c3b4b2062748ed70e689f9f9ffe670e6fa20a071/torch/fx/experimental/shape_prop.py#L7). This is a pattern users will likely find repeatedly useful so it's good to have it first-class in the API. As seen in the test cases, this makes it incredibly easy to do simple analysis and transforms like swapping out individual nodes 2. Add a small `Transformer` class that is an `Interpreter` but provides a `transform` method that yields a GraphModule from interpreting the code   TODO: - [x] Write docstrings - [x] Write docs - [ ] Look for more places this can be used Differential Revision: [D25880330](https://our.internmc.facebook.com/intern/diff/D25880330) [ghstack-poisoned]
1. Codify the interpreter pattern that was used, for ex, in [shape prop](https://github.com/pytorch/pytorch/blob/c3b4b2062748ed70e689f9f9ffe670e6fa20a071/torch/fx/experimental/shape_prop.py#L7). This is a pattern users will likely find repeatedly useful so it's good to have it first-class in the API. As seen in the test cases, this makes it incredibly easy to do simple analysis and transforms like swapping out individual nodes 2. Add a small `Transformer` class that is an `Interpreter` but provides a `transform` method that yields a GraphModule from interpreting the code   TODO: - [x] Write docstrings - [x] Write docs - [ ] Look for more places this can be used Differential Revision: [D25880330](https://our.internmc.facebook.com/intern/diff/D25880330) [ghstack-poisoned]
1. Codify the interpreter pattern that was used, for ex, in [shape prop](https://github.com/pytorch/pytorch/blob/c3b4b2062748ed70e689f9f9ffe670e6fa20a071/torch/fx/experimental/shape_prop.py#L7). This is a pattern users will likely find repeatedly useful so it's good to have it first-class in the API. As seen in the test cases, this makes it incredibly easy to do simple analysis and transforms like swapping out individual nodes 2. Add a small `Transformer` class that is an `Interpreter` but provides a `transform` method that yields a GraphModule from interpreting the code   TODO: - [x] Write docstrings - [x] Write docs - [ ] Look for more places this can be used Differential Revision: [D25880330](https://our.internmc.facebook.com/intern/diff/D25880330) [ghstack-poisoned]
zdevito
approved these changes
Feb 1, 2021
|
@jamesr66a merged this pull request in 609f76f. |
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:
Codify the interpreter pattern that was used, for ex, in shape prop. This is a pattern users will likely find repeatedly useful so it's good to have it first-class in the API. As seen in the test cases, this makes it incredibly easy to do simple analysis and transforms like swapping out individual nodes
Add a small
Transformerclass that is anInterpreterbut provides atransformmethod that yields a GraphModule from interpreting the codeTODO:
Differential Revision: D25880330