feature: slim megabyte-scale dense output maps to a canary and digest before committing the summary - #19
Merged
Conversation
… before committing the summary
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
The ESP32-S3 benchmark firmware dumps a model's entire output tensor. For classifiers that
is O(10) values, but a dense spatial output (the 256x256x8 U-Net segmentation map = 524288
int8) would bloat the tracked
summary.jsonby megabytes. This adds a post-validationslimming step so such a cell can be committed compactly.
Change
tflm-esp32s3/scripts/slim_summary.py: replaces anyoutput_valuesvector longer than4096 elements with a 12-int canary slice plus an
output_digest(len,sha256).Classifier cells are left unchanged.
run_all.sh collect_and_validateAFTERvalidate_accuracy.py(which stillchecks the full device vector against the model reference) and before the summary is
published, so the numerical parity gate is unaffected.
tests/test_slim_summary.py: large output trimmed to canary + digest, small outputunchanged, failure cell unchanged, output-contract preserved, summary roundtrip.
Notes
Full-resolution parity stays reproducible off-summary via
host_parity_unet.py. AU-Net-free summary regenerates byte-identically (classifier cells are untouched). Smoke: a
7.68 MB candidate slims to 666 bytes.