[quant][pt2e][refactor] Remove backend_config from _maybe_insert_input_observers_for_node
#98094
Conversation
…ut_observers_for_node` Summary: The goal is to remove the need to use backend_config when pt2e flow code call this function Test Plan: python test/test_quantization.py TestQuantizeFx Reviewers: Subscribers: Tasks: Tags: [ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/98094
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 1f0dc79: This comment was automatically generated by Dr. CI and updates every 15 minutes. |
…_insert_input_observers_for_node`" Summary: The goal is to remove the need to use backend_config when pt2e flow code call this function Test Plan: python test/test_quantization.py TestQuantizeFx Reviewers: Subscribers: Tasks: Tags: [ghstack-poisoned]
jerryzh168
changed the title
backend_config of _maybe_insert_input_observers_for_nodebackend_config from _maybe_insert_input_observers_for_node
…t_observers_for_node` optional Summary: The goal is to enable pt2e flow code call this function without a `backend_config`, we removed `backend_config` in the non-standalone module branch in this PR so we can reuse this code. The standalone_module branch still needs a backend_config, but that doesn't affect pt2e flow Test Plan: python test/test_quantization.py TestQuantizeFx Reviewers: Subscribers: Tasks: Tags: ghstack-source-id: 7a186e3 Pull Request resolved: #98094
can you share some context on why? |
sure, we don't need this in prepare since validation should be done in quantizer.validate(...) function, so even we have this object (or something similar), it should live in quantizer instead |
| weight_index = node.meta["target_dtype_info"].get("weight_index", None) | ||
| if weight_index is not None and weight_index < len(node.args) and node.args[weight_index] is arg: | ||
| return True | ||
| return node.kwargs.get("weight") is arg |
There was a problem hiding this comment.
I think you can just return False here?
There was a problem hiding this comment.
this might be needed for fx graph mode quantization
…be_insert_input_observers_for_node`" Summary: The goal is to remove the need to use backend_config when pt2e flow code call this function Test Plan: python test/test_quantization.py TestQuantizeFx Reviewers: Subscribers: Tasks: Tags: [ghstack-poisoned]
…be_insert_input_observers_for_node`" Summary: The goal is to remove the need to use backend_config when pt2e flow code call this function Test Plan: python test/test_quantization.py TestQuantizeFx Reviewers: Subscribers: Tasks: Tags: [ghstack-poisoned]
…t_observers_for_node` optional Summary: The goal is to enable pt2e flow code call this function without a `backend_config`, we removed `backend_config` in the non-standalone module branch in this PR so we can reuse this code. The standalone_module branch still needs a backend_config, but that doesn't affect pt2e flow Test Plan: python test/test_quantization.py TestQuantizeFx Reviewers: Subscribers: Tasks: Tags: ghstack-source-id: 11cd27e Pull Request resolved: #98094
|
@pytorchbot merge |
Merge startedYour change will be merged once all checks pass (ETA 0-4 Hours). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
Merge failedReason: 1 jobs have failed, first few of them are: trunk / macos-12-py3-arm64-mps / test (default, 1, 1) Details for Dev Infra teamRaised by workflow job |
|
@pytorchbot merge |
Merge startedYour change will be merged once all checks pass (ETA 0-4 Hours). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
4 participants
Stack from ghstack (oldest at bottom):
backend_configfrom_maybe_insert_input_observers_for_node#98094Summary:
The goal is to remove the need to use backend_config when pt2e flow code call this function
Test Plan:
python test/test_quantization.py TestQuantizeFx
Reviewers:
Subscribers:
Tasks:
Tags: