Only support int8 and quant dtypes for quant operators #11685
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/11685
Note: Links to docs will display an error until the docs builds have been completed. ❌ 1 New Failure, 3 Unrelated FailuresAs of commit b0d9e08 with merge base ec67249 ( NEW FAILURE - The following job has failed:
BROKEN TRUNK - The following jobs failed but were present on the merge base:👉 Rebase onto the `viable/strict` branch to avoid these failures
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
This PR needs a
|
|
@mcr229 @digantdesai I'm running into some issues creating tests. Could you please provide some guidance on some reference files, etc.? |
| @@ -220,9 +225,12 @@ def _check_node_has_valid_dtype(self, node): | |||
| valid_dtypes = { | |||
| torch.float32, | |||
| torch.float16, | |||
| torch.int8, | |||
| torch.qint8, | |||
There was a problem hiding this comment.
what about this guy here?
There was a problem hiding this comment.
That makes sense, I've moved it into the conditional check and added other quant dtypes.
|
@GregoryComer can you point how to run the int8 tests that were failing in the new backend tester? Aside from that as long as existing tests are passing this change should be fine. |
|
rebased and fixed lint |
|
@GregoryComer does this fix your test? |
|
@digantdesai @mcr229 is there anything needed on my end to support this? |
|
@digantdesai has imported this pull request. If you are a Meta employee, you can view this in D79721107. |
|
@digantdesai has imported this pull request. If you are a Meta employee, you can view this in D79721107. |
Summary
Fixes #11581. Check that node is a quant op before including
int8and other quantdtypesinvalid_dtypes.Test plan
Still working on creating tests.