fix(parity): resume at the next pending precision after preemption - #152
Merged
Conversation
Each restart redid every precision stage (~4h of fp32+fp16 decode) because the loop had no completion check. The margin report is the last artifact a stage writes, so its presence on the volume marks the stage durably complete; restarts now skip straight to the pending one.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Why
The 2.1 parity chain spent the day in a preemption loop: Modal restarts
the function with the same input, but the precision loop redid fp32 and
fp16 decode (~4h) on every restart before reaching the pending int8
stage — which then got preempted again.
What
pending_precisions(out_dir, mid, precisions): a stage is completewhen its
*-margins-{precision}.jsonreport exists — that is the lastartifact the stage writes (after the parity block lands in the zip).
parity_modelnow iterates only pending stages; durable stages areskipped, so restarts resume at the next precision.
Evidence
secryst-models (fp32/fp16 reported done repeatedly)