[PyTorch] Deprecate numpy serialization for MTIA #157884
Conversation
Summary: NumPy based tensor rebuilding from serialization has been deprecated by other backends (eg. [XLA](pytorch#137444)). The new flow has CPU storage being constructed with data from the file and then moved to the target backend device. Furthermore, relying on numpy for serialization will fail loudly when torch.load flips weights_only. Reviewed By: andyanwang Differential Revision: D77843238
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/157884
Note: Links to docs will display an error until the docs builds have been completed. ❌ 1 Cancelled Job, 2 Unrelated FailuresAs of commit 010e03f with merge base bc6e066 ( CANCELLED JOB - The following job was cancelled. Please retry:
FLAKY - The following jobs failed but were likely due to flakiness present on trunk:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
|
This pull request was exported from Phabricator. Differential Revision: D77843238 |
nautsimon
added
module: mtia
|
@pytorchbot merge -i (Initiating merge automatically since Phabricator Diff has merged, merging with -i because oss signals were bypassed internally) |
Merge startedYour change will be merged while ignoring the following 3 checks: pull / linux-jammy-py3.9-gcc11 / test (default, 3, 5, linux.2xlarge), pull / linux-jammy-py3.9-gcc11 / test (docs_test, 1, 1, linux.2xlarge), pull / linux-jammy-py3.9-gcc11 / test (distributed, 2, 2, linux.2xlarge) Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
4 participants
Summary:
NumPy based tensor rebuilding from serialization has been deprecated by other backends (eg. XLA). The new flow has CPU storage being constructed with data from the file and then moved to the target backend device.
Furthermore, relying on numpy for serialization will fail loudly when torch.load flips weights_only.
Reviewed By: andyanwang
Differential Revision: D77843238
cc @egienvalue