[Operator] Align CPU randperm_kernel with PyTorch for bit-exact output#78614
Open
zhengshengning wants to merge 3 commits intoPaddlePaddle:developfrom
Open
[Operator] Align CPU randperm_kernel with PyTorch for bit-exact output#78614zhengshengning wants to merge 3 commits intoPaddlePaddle:developfrom
zhengshengning wants to merge 3 commits intoPaddlePaddle:developfrom
Conversation
Replace std::mt19937_64 + std::shuffle with a custom MT19937 engine that matches PyTorch's implementation, ensuring identical randperm results under the same seed. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
|
你的PR提交成功，感谢你对开源项目的贡献! |
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## develop #78614 +/- ##
===========================================
Coverage ? 100.00%
===========================================
Files ? 1
Lines ? 57
Branches ? 0
===========================================
Hits ? 57
Misses ? 0
Partials ? 0 ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
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
5 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
PR Category
Operator Mechanism
PR Types
Improvements
Description
将 CPU 上的
randperm_kernel实现与 PyTorch 对齐，使得在相同 seed 下paddle.randperm与torch.randperm产生 bit-for-bit 一致的输出。改动要点
TorchMT19937Engine类，实现与 PyTorch 完全一致的 Mersenne Twister 随机数引擎（注意：该引擎的行为与std::mt19937/std::mt19937_64不同）。std::mt19937_64+std::shuffle的方案替换为基于 Fisher-Yates shuffle 的实现：n < UINT32_MAX / 20）：使用 32 位随机数的经典 Fisher-Yates shuffle。dev_ctx.GetGenerator()->GetCurrentSeed()获取，移除了原来硬编码seed = 0的逻辑。对齐开启方式：
paddle.set_flags({'FLAGS_use_accuracy_compatible_kernel': 1})
是否引起精度变化
是。在相同 seed 下
randperm的输出序列会发生变化（从与 std 库一致变为与 PyTorch 一致），但 randperm 本身是随机排列算子，不影响模型训练精度。