fix(ci): restore PR #57 + correct speculative-decoding CI failures#61
Closed
solderzzc wants to merge 5 commits into
Closed
fix(ci): restore PR #57 + correct speculative-decoding CI failures#61solderzzc wants to merge 5 commits into
solderzzc wants to merge 5 commits into
Conversation
solderzzc
force-pushed
the
fix/pr57-speculative-ci
branch
from
7b5b14b to
668d474
Compare
Three root causes addressed: 1. Cache key mismatch: speculative jobs used spm-SwiftLM-v2 while build_and_unit_test uses v3, causing a full rebuild on every run. Unified all jobs to v3. 2. Wrong pre-downloaded model: speculative-decoding job pre-fetched Qwen3.5-9B-4bit but test-speculative.sh MAIN_MODEL defaults to Qwen3.5-2B-4bit. The 9B model was never used by the test, and the 2B model was not cached, causing re-download + OOM on 7 GB runner. Pre-download now fetches 2B + 0.8B (matching the test defaults). 4. Fragile metallib compile: replaced python setup.py build_ext with the proven pip install mlx approach already used in build_and_unit_test — eliminates 5-min compile step and pybind11/cmake dependency chain. speculative-decoding-eval retains the 9B model for its heavier memory evaluation (continue-on-error: true already covers OOM). Fixes: CI failure on PR #57
solderzzc
force-pushed
the
fix/pr57-speculative-ci
branch
from
668d474 to
186b8f3
Compare
Member
Author
|
Closing — wrong approach. Fixing the conflict on PR #57 directly. |
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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
This PR re-applies the changes from the failed PR #57 and adds targeted fixes for the three root causes of the CI failure.
What PR #57 contained (cherry-picked cleanly)
huggingface-cli→hfcommand in CIAdditional fixes for the CI failures
1. Cache key mismatch (v2 → v3)
The speculative jobs used
spm-SwiftLM-v2whilebuild_and_unit_testusesv3. This caused a full rebuild on every run instead of sharing the cache. Unified all jobs to v3.2. Wrong pre-downloaded model (9B → 2B)
The
speculative-decodingjob pre-fetchedQwen3.5-9B-4bitbuttest-speculative.shdefaultsMAIN_MODELtoQwen3.5-2B-4bit. The 9B model was never used, the 2B wasn't cached, leading to a re-download at test time + OOM crash on the 7 GB runner (Trace/BPT trap: 5). Pre-download now fetches 2B + 0.8B.3. Fragile metallib compile → simple pip install
Replaced the 5-minute
python setup.py build_ext→ pip-based approach (already proven inbuild_and_unit_test). Removes the pybind11/cmake dependency chain and the fragile glob path.speculative-decoding-evalretains Qwen3.5-9B for its heavier memory evaluation (continue-on-error: truealready covers OOM there).Closes #57