Add jit.ignore to prototype optimizers #2958
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
|
This pull request was exported from Phabricator. Differential Revision: D60943180 |
✅ Deploy Preview for pytorch-fbgemm-docs ready!
To edit notification comments on pull requests, go to your Netlify site configuration. |
|
This pull request was exported from Phabricator. Differential Revision: D60943180 |
spcyppt
added a commit
that referenced
this pull request
Aug 9, 2024
Summary: X-link: facebookresearch/FBGEMM#58 Pull Request resolved: #2958 `torch.compile` doesn't seem to cause errors if we deprecate an optimizer that is no longer used, but `torch.jit.script` will. `torch.jit.script` seems to check and ensure all decision branches are alive. To make prototype optimizers easily deprecated once included in production, we wrap the invoker function with `torch.jit.ignore`. This means that we need to always keep auto-generating the `lookup_{}.py` even the optimizers are deprecated and their backends are removed. [simplified Bento example](https://fburl.com/anp/rbktkl08) Reviewed By: q10 Differential Revision: D60943180
spcyppt
force-pushed
the
export-D60943180
branch
from
74b3327
to
8a5c3fb
Compare
Summary: `torch.compile` doesn't seem to cause errors if we deprecate an optimizer that is no longer used, but `torch.jit.script` will. `torch.jit.script` seems to check and ensure all decision branches are alive. See [simplified Bento example](https://fburl.com/anp/rbktkl08) To make prototype optimizers easily deprecated once included in production, we wrap the invoker function with `torch.jit.ignore`. This means that we need to always keep auto-generating the `lookup_{}.py` even the optimizers are deprecated and their backends are removed. **Usage** Add `"is_prototype_optimizer": True` for the optimizer in `/codegen/genscript/optimizers.py` Example: ``` def ensemble_rowwise_adagrad_optimizer: return { "optimizer": "ensemble_rowwise_adagrad", "is_prototype_optimizer": True, } ``` Reviewed By: q10 Differential Revision: D60943180
|
This pull request was exported from Phabricator. Differential Revision: D60943180 |
spcyppt
force-pushed
the
export-D60943180
branch
from
8a5c3fb
to
bf2ea1e
Compare
|
This pull request has been merged in 9c0aa2a. |
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.
Summary:
torch.compiledoesn't seem to cause errors if we deprecate an optimizer that is no longer used, buttorch.jit.scriptwill.torch.jit.scriptseems to check and ensure all decision branches are alive.To make prototype optimizers easily deprecated once included in production, we wrap the invoker function with
torch.jit.ignore. This means that we need to always keep auto-generating thelookup_{}.pyeven the optimizers are deprecated and their backends are removed.simplified Bento example
Differential Revision: D60943180