-
Notifications
You must be signed in to change notification settings - Fork 537
per_channel_group can't be dynamic #9934
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Conversation
Summary: There are some dynamism issues that arise when checking the semantics of quantize_affine nodes. We avoid them by accounting for free_symbols. Differential Revision: D72488540
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/9934
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit eb44899 with merge base 2d56897 ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
This pull request was exported from Phabricator. Differential Revision: D72488540 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Looks good, we need to be more precise in the code and in the comments/docs about the word dynamic
. We have dynamic shape, dynamic quant, dynamic weights (RHS in bmm in xnnpack), dynamic config. Just easy to assume the wrong one.
@digantdesai yea lol, it can definitely be confusing for folks. I wonder if there are more precise terms we can use to describe the different natures of dynamism. I guess we can add qualifiers like dynamic quantization, dynamic shape, dynamic weights, etc.. |
Differential Revision: D72488540 Pull Request resolved: #9934
Differential Revision: D72488540 Pull Request resolved: pytorch#9934
Summary: There are some dynamism issues that arise when checking the semantics of quantize_affine nodes. We avoid them by accounting for free_symbols.
Differential Revision: D72488540