[STABLE ABI] Port forced_align #4079
Open
Conversation
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
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/audio/4079
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 007452e with merge base 87ff22e ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
5 tasks
pearu
force-pushed
the
pearu/stable-abi-forced_align
branch
from
b3a5a0e to
282bd7f
Compare
samanklesaria
approved these changes
Sep 18, 2025
pearu
commented
Sep 19, 2025
| TORCH_CHECK( | ||
| logProbs.size(1) == at::max(inputLengths).item().toInt(), | ||
| STD_TORCH_CHECK( | ||
| logProbs.size(1) == torchaudio::util::max<int>(inputLengths), |
There was a problem hiding this comment.
This is likely a bug: when inputLenghts is a int64 tensor, int32-casting may produce incorrect result.
pearu
commented
Sep 19, 2025
| C10_CUDA_KERNEL_LAUNCH_CHECK(); | ||
| ++backPtrBufferLen; | ||
| if (backPtrBufferLen == kBackPtrBufferSize || t == T - 1) { | ||
| cpuDataTranferStream.synchronize(); | ||
| // GPU -> GPU copy | ||
| bufferCopy = backPtrBuffer.clone().contiguous(); | ||
| bufferCopy = torchaudio::stable::clone(backPtrBuffer); |
There was a problem hiding this comment.
Nit: use clone(memory_format=contiguous_format) to ensure contiguous bufferCopy.
pearu
commented
Sep 19, 2025
| // template<typename T, size_t N, template <typename U> class PtrTraits = | ||
| // torchaudio::stable::DefaultPtrTraits, typename index_t = int64_t> | ||
| // GenericPackedTensorAccessor<T,N> generic_packed_accessor(const Tensor& t) && | ||
| // = delete; |
pearu
commented
Sep 19, 2025
Comment on lines
+137
to
+139
| // template<typename T, size_t N, template <typename U> class PtrTraits = | ||
| // torchaudio::stable::DefaultPtrTraits> PackedTensorAccessor32<T,N,PtrTraits> | ||
| // packed_accessor32(const Tensor& t) && = delete; |
pearu
commented
Sep 19, 2025
Comment on lines
+150
to
+152
| // template<typename T, size_t N, template <typename U> class PtrTraits = | ||
| // DefaultPtrTraits> PackedTensorAccessor64<T,N,PtrTraits> | ||
| // packed_accessor64(const Tensor& t) && = delete; |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
As in the title.
This PR is marked as Draft as it requires the torch PR stack starting at pytorch/pytorch#161891 (read: CI failures are expected).
Same as #4022 but using C++ torch::stable::Tensor API instead of C shim Tensor API and this PR includes also the GPU port.
PLEASE NOTE THAT THE TORCHAUDIO REPOSITORY IS NO LONGER ACTIVELY MONITORED. You may not get a response. For open discussions, visit https://discuss.pytorch.org/.