[python] Reduce LeRobot DataLoader worker startup time - #9888
Open
XiaoHongbo-Hope wants to merge 4 commits into
Open
XiaoHongbo-Hope wants to merge 4 commits into
XiaoHongbo-Hope wants to merge 4 commits into
Conversation
XiaoHongbo-Hope
force-pushed
the
codex/atomic-s3-metadata-independent
branch
from
September 17, 2026 02:42
826f7b7 to
5417e14
Compare
XiaoHongbo-Hope
changed the title
XiaoHongbo-Hope
marked this pull request as ready for review
September 19, 2026 08:43
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
LeRobot's in-memory episode
Datasetmakesforkserverworker payloads exceed the process pipe capacity, delaying the first batch. Compress the standard pickle for Paimon-created episode metadata; workers restore the originalDatasetstate.ALOHA CPU/local, 8
forkserverworkers: first-batch median 7.83s (3 seeds; unoptimized: 27.72s). Hot throughput stays near 62 samples/s, and sample order matches. PyArrow 19/22: 87 LeRobot tests passed on each.Upstream datasets #8339, included in 5.0.1, serializes only the Arrow table and rebuilds the batch index on load. Versions 4.8.5 and 5.0.0 still serialize both. This PR optimizes the older dependency stack; whether compression remains useful after upgrading to 5.0.1 requires rerunning the worker-startup benchmark.