Repository navigation
Conversation
Authored with assistance from OpenAI Codex.
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Adds a Learning Path for deploying live MobileSAM segmentation on Alif Ensemble E8 with ExecuTorch and Zephyr. It follows the 384×384 four-prompt int8 model through export, firmware memory checks, model and firmware programming, live-camera validation, and troubleshooting.
Review the lessons in navigation order, then the bundled sample patch and OSPI recovery writer. The setup uses released PyTorch 2.13.0, TorchAO 0.18.0, and TorchVision 0.28.0. The Learning Path and its child pages remain marked as drafts pending a fresh firmware build and board run with this dependency setup.
Validation passed: the full site build with Hugo 0.130.0, repository metadata checks, all 19 shell blocks, and a fresh Python 3.12 source installation. The complete four-prompt export passed the unchanged quality gate using the original deployment calibration image, followed by the PTE/JSON tensor-contract checks. The resulting program has no CPU operators and requires 5,898,256 bytes of temporary scratch, within the configured 6,488,064-byte pool. The export lesson now requires an explicit calibration image because the bundled dog image fails this configuration's quality gate.
Authored with assistance from OpenAI Codex.
By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of the Creative Commons Attribution 4.0 International License.