ci: end-to-end reproducibility workflow on free runners - #12
Open
ayushnandi718-dev wants to merge 3 commits into
Open
ci: end-to-end reproducibility workflow on free runners#12ayushnandi718-dev wants to merge 3 commits into
ayushnandi718-dev wants to merge 3 commits into
Conversation
Adds a GitHub Actions workflow that proves the whole pipeline from a clean checkout -- no hardware required: - lint: syntax check + informational ruff run - host-pipeline: prepare a small TinyStories slice, train a tiny ple model on CPU, export model.bin + golden logits, build and run the host verifier (C vs PyTorch golden must PASS) and the perplexity harness, then upload the model artifacts for reuse - firmware-compile: install the ESP32 core + display libs and compile the sketch with arduino-cli, catching C errors without a board Supporting changes: - data/prepare.py: --max-bytes flag so smoke runs use a small slice instead of the full 300MB - src/gen_assets.py: --tok argument so vocab.h can be generated from any tokenizer JSON (not just bpe32768.json)
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Proves the whole pipeline from a clean checkout — no hardware, no cloud credits, and no compute on contributors' machines (everything runs on GitHub-hosted runners):
plemodel on CPU, exportmodel.bin+ golden logits, then build and run the host verifier (C must match the PyTorch golden) and the perplexity harness; model artifacts are uploaded for reuseSupporting changes:
data/prepare.py:--max-bytesflag (smoke runs use a small slice instead of 300MB)src/gen_assets.py:--tokargument sovocab.hcan be generated from any tokenizerThis directly addresses the spirit of issue #7 ("the model does not exist anywhere") by showing the reproducible pipeline end to end, and adds regression protection for firmware changes (#3, #8 and future PRs).