Accumulate activation calibration ranges#2724
Open
kiwigitops wants to merge 1 commit into
Open
Conversation
Signed-off-by: kiwigitops <kiwisclubco@gmail.com>
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Fixes #2704.
Activation calibration was overwriting each tensor's stored
rmin/rmaxbefore comparing the current batch against the previous range. That made the collected range depend on sample order, and a later narrower batch could replace an earlier wider range.This initializes stats only when a tensor is first seen, then accumulates the running min and max for later batches. The same fix is applied to the experimental calibration/debugger paths that record activation ranges.
Tested:
python -m pytest coremltools/test/optimize/test_utils.py::TestUpdateTensorRange coremltools/optimize/coreml/experimental/test_post_training_quantization.py::TestGetActivationStats::test_update_tensor_range_accumulates_running_min_max coremltools/optimize/coreml/experimental/test_post_training_quantization.py::TestGetActivationStats::test_record_intermediate_output_accumulates_running_min_max -qpython -m compileall -q coremltools/optimize/_utils.py coremltools/optimize/coreml/experimental/_post_training_quantization.py coremltools/optimize/coreml/experimental/_model_debugger.py coremltools/test/optimize/test_utils.py coremltools/optimize/coreml/experimental/test_post_training_quantization.pygit diff --check