pytorch/pytorch issues and pull requests

#165548 - [CUDA][cuBLASLt] addmm -- enable 2D bias in the Lt path

Pull Request - State: open - Opened by nikitaved about 1 month ago - 1 comment
Labels: module: cuda, open source, ciflow/trunk, topic: not user facing, matrix multiplication, ciflow/rocm, ciflow/rocm-mi300, ciflow/h100, ciflow/h100-symm-mem, ciflow/b200

#164586 - [CUDA][Muon] bump tolerances for Muon test

Pull Request - State: open - Opened by eqy about 2 months ago
Labels: module: optimizer, module: cuda, module: tests, open source, topic: not user facing, matrix multiplication

#164563 - [Blackwell][Inductor] Numerical mismatches in test_max_autotune.py::TestMaxAutotune::test_non_contiguous_input_mm_plus_mm

Issue - State: open - Opened by Aidyn-A about 2 months ago
Labels: module: cuda, module: inductor, Blackwell

#164480 - [CUDA][Inductor][B200] re-bump tolerances for `test_baddmm` in `test_max_autotune.py`

Pull Request - State: open - Opened by eqy about 2 months ago
Labels: module: cuda, open source, topic: not user facing, Blackwell, ciflow/b200

#164354 - Remove workaround to old CUDA bug

Pull Request - State: open - Opened by pearu about 2 months ago - 4 comments
Labels: module: cuda, module: cpu, open source, release notes: cpp, topic: not user facing

#164049 - [CUDA] fix indexing on large tensor causing nvalid configuration argument

Pull Request - State: closed - Opened by Isalia20 about 2 months ago - 5 comments
Labels: module: cuda, open source, Merged, ciflow/trunk, release notes: cuda, topic: bug fixes, merging

#164048 - [CUDA] indexing on large tensor causes invalid configuration argument

Issue - State: closed - Opened by Isalia20 about 2 months ago
Labels: module: cuda

#163664 - [BE] Add Linux aarch64 CUDA install and test to validation framework

Issue - State: closed - Opened by atalman about 2 months ago
Labels: module: binaries, module: cuda, triaged, better-engineering, topic: binaries

#163658 - [CI] Downgrading CUDA driver results: No devices were found

Issue - State: closed - Opened by atalman about 2 months ago - 1 comment
Labels: module: cuda, module: ci, triaged, module: third_party, has workaround

#163581 - [cuDNN][Convolution] Disable cuDNN for 3D convolutions with kernel size != 1 for cuDNN 9.8+

Pull Request - State: closed - Opened by eqy about 2 months ago - 7 comments
Labels: module: cudnn, module: cuda, module: cpu, module: convolution, triaged, open source, ciflow/trunk, topic: bug fixes, release notes: cudnn, module: inductor, ciflow/inductor, merging

#163342 - [CD] - Manywheel CUDA builds failing since Sept 18

Issue - State: closed - Opened by robert-hardwick 2 months ago - 5 comments
Labels: high priority, triage review, module: binaries, module: cuda, triaged, module: regression

#163299 - [CUDA] Cleanup persistent cuBLASLt workspaces before compile-regions test

Pull Request - State: closed - Opened by eqy 2 months ago - 4 comments
Labels: module: cuda, module: tests, triaged, module: cublas, open source, Merged, ciflow/trunk, topic: not user facing, module: CUDACachingAllocator, matrix multiplication, merging

#163070 - [TEST][CUDA] Use proper dtype in test_cuda_tensor_pow_scalar_tensor_cuda

Pull Request - State: closed - Opened by Aidyn-A 2 months ago - 3 comments
Labels: module: cuda, open source, Merged, ciflow/trunk, topic: not user facing, merging

#163001 - [CUDA13][CUDA Graphs] Fix `test_graph_external_wait_and_record` to account for minimum supported compute-capability

Pull Request - State: open - Opened by eqy 2 months ago
Labels: module: cuda, open source, module: cuda graphs, topic: not user facing

#162995 - [CUDA] fix shared memory race in `reduce_kernel`

Pull Request - State: open - Opened by eqy 2 months ago
Labels: module: cuda, open source, module: reductions, topic: not user facing

#162626 - compile_kernel: Handle python floats as c double

Pull Request - State: closed - Opened by msaroufim 2 months ago - 17 comments
Labels: module: cuda, ciflow/trunk, release notes: cuda, topic: bug fixes, merging

#162578 - [BUG] Fix nonzero_static crash on CUDA when the input is a empty tensor

Pull Request - State: closed - Opened by can-gaa-hou 2 months ago - 10 comments
Labels: module: cuda, triaged, open source, Merged, ciflow/trunk, release notes: cuda, topic: bug fixes, merging

#162544 - Build fbgemm_gpu for TORCH_CUDA_ARCH_LIST=10.0 and CUDA 12.8 and 12.9

Pull Request - State: closed - Opened by danielvegamyhre 2 months ago - 4 comments
Labels: module: cuda, Merged, ciflow/trunk, topic: not user facing, merging

#162429 - [CUDA][Distributed][CI] PyTorch Encounters "CUDA driver error: invalid argument" Failures with Distributed CUDASymmetricMemory Tests on B200

Issue - State: closed - Opened by nWEIdia 2 months ago - 5 comments
Labels: oncall: distributed, module: cuda

#162333 - [CD] Windows CUDA 13.0 binaries : Windows fatal exception: access violation

Issue - State: closed - Opened by atalman 2 months ago - 9 comments
Labels: high priority, module: binaries, module: windows, module: cuda, oncall: releng, triaged

#162322 - [CUDA 13][cuDNN][Windows] Roll back cuDNN upgrade from 9.13 to 9.12 on Windows

Pull Request - State: closed - Opened by eqy 2 months ago - 3 comments
Labels: module: windows, module: cudnn, module: cuda, open source, Merged, ciflow/trunk, topic: not user facing, merging

#162203 - Update round size with 1 division behavior

Pull Request - State: open - Opened by morrison-turnansky 3 months ago - 5 comments
Labels: oncall: distributed, module: cuda, module: cpu, triaged, open source, NNC, release notes: quantization, topic: not user facing, module: inductor, module: dynamo, release notes: distributed (checkpoint), module: compiled autograd, release notes: inductor (aoti)

#162186 - [CUDA][CUDAGraph] Reduce capture overhead in CUDA Graph memory reuse

Pull Request - State: closed - Opened by eee4017 3 months ago - 11 comments
Labels: module: cuda, open source, Merged, module: cuda graphs, ciflow/trunk, topic: not user facing, merging, ciflow/rocm

#162185 - [B200][NVFP4] Fix argument passing in `test_blockwise_mxfp8_nvfp4_mxfp4_numerics_`

Issue - State: closed - Opened by eqy 3 months ago - 3 comments
Labels: module: cuda, open source, Merged, ciflow/trunk, topic: not user facing, merging, module: floatx (formerly float8), ciflow/h100

#162180 - [B200][MXFP8] Fix regex in `test_blockwise_mxfp8_nvfp4_error_messages_recipe_mxfp8_cuda`

Pull Request - State: closed - Opened by eqy 3 months ago - 3 comments
Labels: module: cuda, open source, Merged, ciflow/trunk, topic: not user facing, merging, module: floatx (formerly float8)

#162073 - [cuDNN][SDPA] Enable cuDNN SDPA by default for SM 9.0, SM 10.0

Pull Request - State: open - Opened by eqy 3 months ago
Labels: module: cudnn, module: cuda, open source, topic: not user facing, module: sdpa

#161884 - IPC in the ExpandableSegment can not correctly match the handle from the map ipcMemHandle_to_devptr.

Issue - State: closed - Opened by mengph 3 months ago - 1 comment
Labels: module: cuda, triaged

#161822 - [CUDA][cuBLAS] #125888 introduces measurable CPU overhead for matmuls

Issue - State: closed - Opened by eqy 3 months ago - 4 comments
Labels: module: performance, module: cuda, triaged, module: python frontend

#161789 - "cudaErrorIllegalAddress" in unique_consecutive (with return_inverse and RMM)

Issue - State: closed - Opened by nboeschen 3 months ago - 1 comment
Labels: module: crash, module: cuda, triaged

#161749 - [cuBLAS] update cuBLAS determinism docs, remove workspace requirement checks

Pull Request - State: closed - Opened by eqy 3 months ago - 8 comments
Labels: module: cuda, triaged, module: cublas, module: determinism, open source, Merged, ciflow/trunk, release notes: cuda, merging, ciflow/h100

#161649 - Use vectorized stores for all dtypes in cat

Pull Request - State: closed - Opened by ngimel 3 months ago - 31 comments
Labels: module: cuda, Merged, Reverted, ciflow/trunk, release notes: cuda, ci-no-td

#161581 - DISABLED test_autocast_ignored_types (main.TestCudaAutocast)

Issue - State: open - Opened by pytorch-bot[bot] 3 months ago - 3 comments
Labels: module: cuda, module: rocm, triaged, module: flaky-tests, skipped

#161481 - About torch.UntypedStorage._new_shared_cuda

Issue - State: closed - Opened by Schnabel-8 3 months ago - 1 comment
Labels: module: multiprocessing, module: cuda, triaged

#161434 - [cuDNN][SDPA][Nested Tensor] add forward/backward caching support for cuDNN SDPA Nested tensor/varlen

Pull Request - State: open - Opened by eqy 3 months ago
Labels: module: cudnn, module: cuda, open source, module: nestedtensor, topic: not user facing, module: sdpa

#161399 - [ATen][CUDA] Add family conditional for CUTLASS matmuls

Pull Request - State: closed - Opened by Aidyn-A 3 months ago - 3 comments
Labels: module: cuda, triaged, open source, release notes: cuda, matrix multiplication, module: floatx (formerly float8), module: core aten

#161380 - CUDA 13 -- sm_120 -- Nvidia 5090 -- ptxas warning : Value of threads …

Pull Request - State: closed - Opened by DrStone71 3 months ago - 15 comments
Labels: module: cuda, triaged, open source, Merged, ciflow/trunk, topic: bug fixes, topic: not user facing, merging

#161305 - [cuBLASLt][FP8] `cuBLASLt` appears to support float8 rowwise-scaling on H100

Pull Request - State: open - Opened by eqy 3 months ago - 13 comments
Labels: module: cuda, module: cublas, open source, Merged, Reverted, ciflow/trunk, topic: not user facing, module: floatx (formerly float8), ci-no-td, ciflow/rocm-mi300

#161177 - [cuDNN][convolution] remove redundant conv3d 64bit test

Pull Request - State: closed - Opened by eqy 3 months ago - 15 comments
Labels: module: cudnn, module: cuda, open source, Merged, module: tf32, ciflow/trunk, topic: not user facing, merging

#161139 - `roundup_power2_divisions` ignores the value of 1

Issue - State: closed - Opened by nick-griaznov 3 months ago
Labels: module: cuda, triaged, module: CUDACachingAllocator

#161107 - [CUDA] Don't import hipify modules on CUDA

Pull Request - State: open - Opened by eqy 3 months ago
Labels: module: cuda, open source, topic: not user facing

#161046 - torch.ones(1, device=torch.cuda.current_device()) CUDA error: operation not supported

Issue - State: closed - Opened by EvilCalf 3 months ago - 2 comments
Labels: needs reproduction, module: cuda, triaged

#160992 - fix-unpin-memory-tensor-param

Pull Request - State: closed - Opened by ghostspiders 3 months ago - 9 comments
Labels: oncall: distributed, module: cuda, triaged, open source, Merged, ciflow/trunk, release notes: distributed (checkpoint), merging

#160960 - compile error: 'nvmlDeviceGetGpuFabricInfoV' was not declared in this scope

Issue - State: closed - Opened by can-gaa-hou 3 months ago - 1 comment
Labels: module: build, module: cuda, triaged, actionable

#160719 - DISABLED test_graph_partition_cpu_scalar_multiple (main.CudaGraphTreeTests)

Issue - State: closed - Opened by pytorch-bot[bot] 3 months ago - 8 comments
Labels: module: cuda, triaged, module: flaky-tests, skipped

#160693 - [FP8][cuBLAS][SM100] cuBLAS doesn't support rowwise-scaling on `sm100`

Pull Request - State: open - Opened by eqy 3 months ago
Labels: module: cuda, module: cublas, open source, topic: not user facing, matrix multiplication, module: floatx (formerly float8)

#160598 - DISABLED test_autocast_custom_enabled (main.TestCudaAutocast)

Issue - State: closed - Opened by pytorch-bot[bot] 3 months ago - 6 comments
Labels: module: cuda, module: rocm, triaged, module: flaky-tests, skipped

#160554 - [ATen][CUDA] Use new CCCL API in v2.8

Pull Request - State: closed - Opened by Aidyn-A 3 months ago - 3 comments
Labels: module: cuda, open source, ciflow/trunk, release notes: cuda, topic: not user facing, merging, module: core aten

#160551 - DISABLED test_autocast_custom_cast_inputs (main.TestCudaAutocast)

Issue - State: closed - Opened by pytorch-bot[bot] 3 months ago - 6 comments
Labels: module: cuda, module: rocm, triaged, module: flaky-tests, skipped

#160507 - DISABLED test_autocast_checkpointing (main.TestCudaAutocast)

Issue - State: closed - Opened by pytorch-bot[bot] 3 months ago - 9 comments
Labels: module: cuda, module: rocm, triaged, module: flaky-tests, skipped

#160385 - TTGIR error for FlexAttention on B200

Issue - State: closed - Opened by drisspg 3 months ago - 2 comments
Labels: module: cuda, triaged, oncall: pt2, upstream triton, module: higher order operators, module: pt2-dispatcher, module: flex attention, Blackwell

#160192 - Flex Attention heuristics: a Blackwell config

Pull Request - State: open - Opened by Aidyn-A 3 months ago
Labels: module: cuda, topic: not user facing, module: inductor, module: flex attention

#159939 - [Test][Easy] Use float16 dtype in test_sort_large

Pull Request - State: closed - Opened by Aidyn-A 4 months ago - 7 comments
Labels: module: cuda, open source, ciflow/trunk, topic: not user facing, merging

#159892 - [RFC] CUDAPluggableAllocator receives malloc request of size zero.

Issue - State: closed - Opened by siyuanchai1999 4 months ago - 4 comments
Labels: module: cuda, triaged

#159802 - DISABLED test_pin_memory_no_cuda (main.TestDictDataLoader)

Issue - State: closed - Opened by izaitsevfb 4 months ago - 2 comments
Labels: module: dataloader, module: cuda, triaged, skipped

#159689 - [CUDA] Skip pynvml test on platforms that don't have complete support

Pull Request - State: open - Opened by eqy 4 months ago
Labels: module: cuda, open source, topic: not user facing

#159672 - [CUDA] Add some more missing `@serialTest` decorators

Pull Request - State: closed - Opened by eqy 4 months ago - 6 comments
Labels: module: cuda, open source, Merged, ciflow/trunk, topic: not user facing, merging

#159663 - DISABLED test_mempool_empty_cache_inactive (main.TestMemPool)

Issue - State: open - Opened by pytorch-bot[bot] 4 months ago - 2 comments
Labels: module: cuda, triaged, module: flaky-tests, skipped

#159466 - [ATen][CUDA][cuFFT] Guard against deprecated error codes

Pull Request - State: closed - Opened by Aidyn-A 4 months ago - 3 comments
Labels: module: cuda, triaged, open source, module: fft, Merged, ciflow/trunk, topic: not user facing, merging, module: core aten

#159446 - not support nvidia 5060

Issue - State: closed - Opened by tuqizhao 4 months ago - 9 comments
Labels: module: windows, module: cuda, triaged

#159309 - torch.nn.InstanceNorm2d CPU/GPU Inconsistency

Issue - State: closed - Opened by ChaitanyaRS06 4 months ago - 1 comment
Labels: module: cuda

#159305 - [CUDA][CUDA Graphs] Move cuda graphs test to subprocess to avoid polluting mempool tests

Pull Request - State: open - Opened by eqy 4 months ago
Labels: module: cuda, open source, module: cuda graphs, topic: not user facing

#159286 - DISABLED test_graph_memory_stats_and_use_result_after_destroy_graph (main.TestCuda)

Issue - State: open - Opened by pytorch-bot[bot] 4 months ago
Labels: module: cuda, triaged, module: flaky-tests, skipped

#159285 - DISABLED test_cuda_kernel_loop_overflow_large (main.TestCuda)

Issue - State: open - Opened by pytorch-bot[bot] 4 months ago
Labels: module: cuda, triaged, module: flaky-tests, skipped

#159284 - DISABLED test_graph_memory_stats_and_use_result_after_destroy_graph (main.TestCuda)

Issue - State: closed - Opened by pytorch-bot[bot] 4 months ago - 1 comment
Labels: module: cuda, triaged, module: flaky-tests, skipped

#159283 - DISABLED test_cuda_kernel_loop_overflow_large (main.TestCuda)

Issue - State: open - Opened by pytorch-bot[bot] 4 months ago
Labels: module: cuda, triaged, module: flaky-tests, skipped

#159271 - [CUDA] Add `serialTest` decorator to `largeTensorTest` in `test_cuda.py`

Pull Request - State: closed - Opened by eqy 4 months ago - 9 comments
Labels: module: cuda, triaged, module: 64-bit, open source, Merged, ciflow/trunk, topic: not user facing, merging

#159265 - Unpin dependency version when possible

Issue - State: open - Opened by malfet 4 months ago
Labels: module: binaries, module: cuda

#159113 - DISABLED test_graph_two_successive (main.TestCuda)

Issue - State: closed - Opened by pytorch-bot[bot] 4 months ago - 2 comments
Labels: module: cuda, triaged, module: flaky-tests, skipped

#159104 - [CUDA] Add experimental green context support for SM carveout

Pull Request - State: closed - Opened by eqy 4 months ago - 17 comments
Labels: module: cuda, triaged, open source, Merged, Reverted, ciflow/trunk, topic: not user facing, matrix multiplication, merging, ci-no-td, lint-all-files

#159068 - DISABLED test_cuda_kernel_loop_overflow (main.TestCuda)

Issue - State: closed - Opened by pytorch-bot[bot] 4 months ago - 1 comment
Labels: module: cuda, triaged, module: flaky-tests, skipped

#159038 - DISABLED test_cuda_memory_leak_detection_propagates_errors (main.TestCuda)

Issue - State: closed - Opened by pytorch-bot[bot] 4 months ago - 1 comment
Labels: module: cuda, triaged, module: flaky-tests, skipped

#158994 - [CUDA] Fix missing `__syncthreads` in MultiMarginLoss backward

Pull Request - State: closed - Opened by eqy 4 months ago - 3 comments
Labels: module: loss, module: cuda, open source, Merged, ciflow/trunk, release notes: nn, topic: bug fixes, merging

#158981 - [conv][cuDNN][64-bit indexing] reduce memory usage of depthwise conv 64-bit indexing test

Pull Request - State: closed - Opened by eqy 4 months ago - 4 comments
Labels: module: cudnn, module: cuda, module: convolution, triaged, module: 64-bit, open source, Merged, ciflow/trunk, topic: not user facing, merging

#158921 - Is it necessary to add a __syncthreads in the MultiMarginLoss_backward_kernel of file aten/src/ATen/native/cuda/MultiMarginLoss.cu

Issue - State: closed - Opened by zzppq 4 months ago - 1 comment
Labels: module: cuda, triaged, module: correctness (silent)

#158764 - DISABLED test_allocate_in_thread_to_pool (main.TestBlockStateAbsorption)

Issue - State: open - Opened by pytorch-bot[bot] 4 months ago - 10 comments
Labels: module: cuda, triaged, module: flaky-tests, skipped

#158494 - [B200] Fix flex-attention heuristic for `test_tma_with_customer_kernel_options_cuda`

Pull Request - State: closed - Opened by eqy 4 months ago - 3 comments
Labels: module: cuda, open source, Merged, ciflow/trunk, topic: not user facing, module: inductor, ciflow/inductor, merging, module: flex attention

#158484 - [Fix] Rework CUDA error explanation framework to be less destructive …

Pull Request - State: closed - Opened by Raymo111 4 months ago - 6 comments
Labels: module: cuda, ciflow/trunk, topic: not user facing, merging

#158395 - Add framework for explanations for common CUDA errors

Pull Request - State: closed - Opened by Raymo111 4 months ago - 3 comments
Labels: module: cuda, Merged, ciflow/trunk, release notes: cuda, merging

#158334 - DISABLED test_live_outputs_multiple_graphs (main.CudaGraphTreeTests)

Issue - State: open - Opened by pytorch-bot[bot] 4 months ago
Labels: module: cuda, module: rocm, triaged, module: flaky-tests, skipped

#158301 - Add warning about removed sm50 and sm60 arches

Pull Request - State: open - Opened by atalman 4 months ago - 1 comment
Labels: module: cuda, topic: not user facing

#158172 - `torch.fmin` has inconsistent overflow behavior on CPU and GPU

Issue - State: open - Opened by jiren-the-gray 4 months ago - 1 comment
Labels: module: cuda, triaged, module: edge cases

#158122 - [RFC] A Distributed CUDA Unified Memory Backend for PyTorch

Issue - State: open - Opened by matthewdcong 4 months ago
Labels: oncall: distributed, module: cuda, triaged, module: PrivateUse1

#158037 - Support DeepSeek-style blockwise scaling scaled-mm for fp8 on Hopper+

Pull Request - State: closed - Opened by lw 4 months ago - 23 comments
Labels: module: cuda, Merged, Reverted, ciflow/trunk, topic: not user facing, merging, module: floatx (formerly float8), ciflow/rocm, ci-no-td

#157999 - [CUDA] Support family-conditional compute capabilies in `TORCH_CUDA_ARCH_LIST`

Pull Request - State: closed - Opened by eqy 4 months ago - 3 comments
Labels: module: build, module: cuda, open source, ciflow/trunk, topic: not user facing, topic: build, merging

#157901 - DISABLED test_graph_partition_reorder_custom_op_with_no_dependency1 (main.CudaGraphTreeTests)

Issue - State: open - Opened by pytorch-bot[bot] 4 months ago - 3 comments
Labels: module: cuda, module: rocm, triaged, module: flaky-tests, skipped

#157791 - [BE]: Reduce binary size 40% using aggressive fatbin compression.

Pull Request - State: closed - Opened by Skylion007 4 months ago - 35 comments
Labels: module: cuda, triaged, open source, better-engineering, Merged, Reverted, Stale, ciflow/trunk, release notes: cuda, release notes: releng, release notes: build, ciflow/binaries_wheel, ciflow/binaries_libtorch, ci-no-td

#157791 - [BE]: Reduce binary size 40% using aggressive fatbin compression.

Pull Request - State: open - Opened by Skylion007 4 months ago - 32 comments
Labels: module: cuda, open source, better-engineering, Merged, Reverted, ciflow/trunk, release notes: cuda, release notes: releng, release notes: build, ciflow/binaries_wheel, ciflow/binaries_libtorch, ci-no-td

#157761 - DISABLED test_graph_partition_reorder_cpu_and_gpu (main.CudaGraphTreeTests)

Issue - State: closed - Opened by pytorch-bot[bot] 4 months ago - 6 comments
Labels: module: cuda, triaged, module: flaky-tests, skipped

#157725 - DISABLED test_resnet (main.TestBlockStateAbsorption)

Issue - State: open - Opened by pytorch-bot[bot] 4 months ago - 9 comments
Labels: module: cuda, triaged, module: flaky-tests, skipped

#157723 - DISABLED test_graph_partition_forward_with_skipped_cudagraphed_backward (main.CudaGraphTreeTests)

Issue - State: closed - Opened by pytorch-bot[bot] 4 months ago - 8 comments
Labels: module: cuda, triaged, module: flaky-tests, skipped

#157711 - [Easy][Profiler] Fix pattern matcher of profiler

Pull Request - State: closed - Opened by Aidyn-A 5 months ago - 14 comments
Labels: module: cuda, triaged, open source, oncall: profiler, Merged, ciflow/trunk, topic: not user facing, merging

#157668 - NCCL error caused due to use of NVLS in torch 2.7.1-cu128 on aarch64 gb200 cluster

Issue - State: open - Opened by mihirp1998 5 months ago
Labels: triage review, module: cuda, module: nccl, module: arm

#157643 - DISABLED test_graph_partition_forward_backward_not_called (main.CudaGraphTreeTests)

Issue - State: closed - Opened by pytorch-bot[bot] 5 months ago - 6 comments
Labels: module: cuda, triaged, module: flaky-tests, skipped

#157616 - DISABLED test_graph_partition_forward_backward (main.CudaGraphTreeTests)

Issue - State: closed - Opened by pytorch-bot[bot] 5 months ago - 6 comments
Labels: module: cuda, triaged, module: flaky-tests, skipped

#157586 - [CUDA][NVTX] use `pytorch` nvtx domain for pytorch ranges

Pull Request - State: closed - Opened by eqy 5 months ago - 3 comments
Labels: module: cuda, triaged, open source, Stale, topic: not user facing

#157535 - Error in Qwen inference: NVML_SUCCESS == r INTERNAL ASSERT FAILED at "/pytorch/c10/cuda/CUDACachingAllocator.cpp

Issue - State: closed - Opened by dineshsoudagar 5 months ago - 3 comments
Labels: high priority, needs reproduction, module: cuda, module: error checking, triaged

GitHub / pytorch/pytorch issues and pull requests