Qwen3 with Vulkan Backend + KV Cache Fails Export #12373
Description
🐛 Describe the bug
With the following configs:
base:
metadata: '{"get_bos_id": 151644, "get_eos_ids":[151645]}'
model:
use_kv_cache: True
use_sdpa_with_kv_cache: True
dtype_override: fp32
enable_dynamic_shape: False
quantization:
qmode: vulkan_4w
backend:
vulkan:
enabled: True
Running:
python -m extension.llm.export.export_llm --config configs.yaml +base.model_class="qwen3_0_6b" +base.params="examples/models/qwen3/config/0_6b_config.json" +export.output_name="qwen3_0_6b.pte"
I get the below error:
...
[2025-07-10 15:15:01,004][root][INFO] - [Vulkan Partitioner] Due to [no operator implementation], skipping %aten_alias_copy_default_1 : [num_users=3] = call_function[target=executorch.exir.dialects.edge._ops.aten.alias_copy.default](args = (%aten_mul_tensor_8,), kwargs = {})
[2025-07-10 15:15:01,004][root][INFO] - [Vulkan Partitioner] Due to [no operator implementation], skipping %aten_alias_copy_default : [num_users=3] = call_function[target=executorch.exir.dialects.edge._ops.aten.alias_copy.default](args = (%aten_mul_tensor_5,), kwargs = {})
[2025-07-10 15:15:01,174][root][INFO] - [Vulkan Partitioner] Due to [no operator implementation], skipping %aten_index_tensor_1 : [num_users=28] = call_function[target=executorch.exir.dialects.edge._ops.aten.index.Tensor](args = (%b_rope_freqs_sin, [%attn_options_input_pos]), kwargs = {})
[2025-07-10 15:15:01,174][root][INFO] - [Vulkan Partitioner] Due to [no operator implementation], skipping %aten_index_tensor : [num_users=28] = call_function[target=executorch.exir.dialects.edge._ops.aten.index.Tensor](args = (%b_rope_freqs_cos, [%attn_options_input_pos]), kwargs = {})
[2025-07-10 15:15:01,174][root][INFO] - [Vulkan Partitioner] Due to [op requires buffers that exceed the buffer limit (134217728)], skipping %aten_embedding_default : [num_users=3] = call_function[target=executorch.exir.dialects.edge._ops.aten.embedding.default](args = (%p_tok_embeddings_weight, %tokens), kwargs = {})
[2025-07-10 15:15:01,339][root][INFO] - Found 29 Vulkan subgraphs to be partitioned.
Error executing job with overrides: ['+base.model_class=qwen3_0_6b', '+base.params=examples/models/qwen3/config/0_6b_config.json', '+export.output_name=qwen3_0_6b.pte']
Traceback (most recent call last):
File "/home/a.dean1/dev/executorchMain/executorch/extension/llm/export/export_llm.py", line 88, in hydra_main
export_llama(llm_config_obj)
File "/home/a.dean1/miniconda3/envs/executorch/lib/python3.10/site-packages/executorch/examples/models/llama/export_llama_lib.py", line 635, in export_llama
builder = _export_llama(llm_config)
File "/home/a.dean1/miniconda3/envs/executorch/lib/python3.10/site-packages/executorch/examples/models/llama/export_llama_lib.py", line 1081, in _export_llama
builder = _to_edge_and_lower_llama(
File "/home/a.dean1/miniconda3/envs/executorch/lib/python3.10/site-packages/executorch/examples/models/llama/export_llama_lib.py", line 1032, in _to_edge_and_lower_llama
builder = builder_exported_to_edge.to_backend(partitioners)
File "/home/a.dean1/miniconda3/envs/executorch/lib/python3.10/site-packages/executorch/extension/llm/export/builder.py", line 455, in to_backend
self.edge_manager = self.edge_manager.to_backend(partitioner)
File "/home/a.dean1/miniconda3/envs/executorch/lib/python3.10/site-packages/executorch/exir/program/_program.py", line 114, in wrapper
return func(self, *args, **kwargs)
File "/home/a.dean1/miniconda3/envs/executorch/lib/python3.10/site-packages/executorch/exir/program/_program.py", line 1531, in to_backend
new_edge_programs = to_backend(method_to_programs_and_partitioners)
File "/home/a.dean1/miniconda3/envs/executorch/lib/python3.10/functools.py", line 889, in wrapper
return dispatch(args[0].__class__)(*args, **kw)
File "/home/a.dean1/miniconda3/envs/executorch/lib/python3.10/site-packages/executorch/exir/backend/backend_api.py", line 734, in _
backend_id_to_call_submodule_nodes = _create_partitions(
File "/home/a.dean1/miniconda3/envs/executorch/lib/python3.10/site-packages/executorch/exir/backend/backend_api.py", line 536, in _create_partitions
backend_id_to_call_submodules = _create_partitions_in_graph_module(
File "/home/a.dean1/miniconda3/envs/executorch/lib/python3.10/site-packages/executorch/exir/backend/backend_api.py", line 485, in _create_partitions_in_graph_module
) = create_exported_program_from_submodule(
File "/home/a.dean1/miniconda3/envs/executorch/lib/python3.10/site-packages/executorch/exir/lowered_backend_module.py", line 723, in create_exported_program_from_submodule
ExportedProgram(
File "/home/a.dean1/miniconda3/envs/executorch/lib/python3.10/site-packages/torch/export/exported_program.py", line 1072, in __init__
self.validate()
File "/home/a.dean1/miniconda3/envs/executorch/lib/python3.10/site-packages/torch/export/exported_program.py", line 1628, in validate
self._validate()
File "/home/a.dean1/miniconda3/envs/executorch/lib/python3.10/site-packages/torch/export/exported_program.py", line 1637, in _validate
v().check(self)
File "/home/a.dean1/miniconda3/envs/executorch/lib/python3.10/site-packages/torch/_export/verifier.py", line 185, in check
_verify_exported_program_signature(ep)
File "/home/a.dean1/miniconda3/envs/executorch/lib/python3.10/site-packages/torch/_export/verifier.py", line 485, in _verify_exported_program_signature
raise SpecViolationError(
torch._export.verifier.SpecViolationError: Mutation node _local_scalar_dense_52 is neither a buffer nor a user input. Buffers to mutate: {'getitem_53': 'layers.26.attention.kv_cache.v_cache', 'getitem_52': 'layers.26.attention.kv_cache.k_cache'}, User inputs to mutate: {}
Set the environment variable HYDRA_FULL_ERROR=1 for a complete stack trace.
This is error is very similar to this issue: #6373
But it looks like the solution to it is outdated/doesn't apply to this model.
Let me know if you need more data to reproduce! Thanks :)
Versions
PyTorch version: 2.9.0.dev20250706+cpu
Is debug build: False
CUDA used to build PyTorch: None
ROCM used to build PyTorch: N/A
OS: Ubuntu 20.04.6 LTS (x86_64)
GCC version: (Ubuntu 9.4.0-1ubuntu1~20.04.2) 9.4.0
Clang version: Could not collect
CMake version: version 3.30.4
Libc version: glibc-2.31
Python version: 3.10.17 | packaged by conda-forge | (main, Apr 10 2025, 22:19:12) [GCC 13.3.0] (64-bit runtime)
Python platform: Linux-5.15.0-139-generic-x86_64-with-glibc2.31
Is CUDA available: False
CUDA runtime version: No CUDA
CUDA_MODULE_LOADING set to: N/A
GPU models and configuration: No CUDA
Nvidia driver version: No CUDA
cuDNN version: No CUDA
Is XPU available: False
HIP runtime version: N/A
MIOpen runtime version: N/A
Is XNNPACK available: True
CPU:
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Byte Order: Little Endian
Address sizes: 46 bits physical, 48 bits virtual
CPU(s): 48
On-line CPU(s) list: 0-47
Thread(s) per core: 2
Core(s) per socket: 24
Socket(s): 1
NUMA node(s): 1
Vendor ID: GenuineIntel
CPU family: 6
Model: 85
Model name: Intel(R) Xeon(R) Gold 5220R CPU @ 2.20GHz
Stepping: 7
CPU MHz: 2200.000
CPU max MHz: 4000.0000
CPU min MHz: 1000.0000
BogoMIPS: 4400.00
Virtualization: VT-x
L1d cache: 768 KiB
L1i cache: 768 KiB
L2 cache: 24 MiB
L3 cache: 35.8 MiB
NUMA node0 CPU(s): 0-47
Vulnerability Gather data sampling: Mitigation; Microcode
Vulnerability Itlb multihit: KVM: Mitigation: VMX disabled
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Not affected
Vulnerability Mmio stale data: Mitigation; Clear CPU buffers; SMT vulnerable
Vulnerability Reg file data sampling: Not affected
Vulnerability Retbleed: Mitigation; Enhanced IBRS
Vulnerability Spec rstack overflow: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl and seccomp
Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2: Mitigation; Enhanced / Automatic IBRS; IBPB conditional; RSB filling; PBRSB-eIBRS SW sequence; BHI SW loop, KVM SW loop
Vulnerability Srbds: Not affected
Vulnerability Tsx async abort: Mitigation; TSX disabled
Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb cat_l3 cdp_l3 invpcid_single intel_ppin ssbd mba ibrs ibpb stibp ibrs_enhanced tpr_shadow vnmi flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid cqm mpx rdt_a avx512f avx512dq rdseed adx smap clflushopt clwb intel_pt avx512cd avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local dtherm ida arat pln pts hwp hwp_act_window hwp_epp hwp_pkg_req pku ospke avx512_vnni md_clear flush_l1d arch_capabilities
Versions of relevant libraries:
[pip3] executorch==0.8.0a0+7e33791
[pip3] numpy==2.2.5
[pip3] nvidia-cublas-cu12==12.6.4.1
[pip3] nvidia-cuda-cupti-cu12==12.6.80
[pip3] nvidia-cuda-nvrtc-cu12==12.6.77
[pip3] nvidia-cuda-runtime-cu12==12.6.77
[pip3] nvidia-cudnn-cu12==9.5.1.17
[pip3] nvidia-cufft-cu12==11.3.0.4
[pip3] nvidia-curand-cu12==10.3.7.77
[pip3] nvidia-cusolver-cu12==11.7.1.2
[pip3] nvidia-cusparse-cu12==12.5.4.2
[pip3] nvidia-cusparselt-cu12==0.6.3
[pip3] nvidia-nccl-cu12==2.26.2
[pip3] nvidia-nvjitlink-cu12==12.6.85
[pip3] nvidia-nvtx-cu12==12.6.77
[pip3] pytorch_tokenizers==0.1.0
[pip3] torch==2.9.0.dev20250706+cpu
[pip3] torchao==0.12.0+gitbc68b11f
[pip3] torchaudio==2.8.0.dev20250706+cpu
[pip3] torchdata==0.11.0
[pip3] torchsr==1.0.4
[pip3] torchtune==0.6.1
[pip3] torchvision==0.24.0.dev20250706+cpu
[pip3] triton==3.3.0
[conda] executorch 0.8.0a0+7e33791 pypi_0 pypi
[conda] numpy 2.2.5 pypi_0 pypi
[conda] nvidia-cublas-cu12 12.6.4.1 pypi_0 pypi
[conda] nvidia-cuda-cupti-cu12 12.6.80 pypi_0 pypi
[conda] nvidia-cuda-nvrtc-cu12 12.6.77 pypi_0 pypi
[conda] nvidia-cuda-runtime-cu12 12.6.77 pypi_0 pypi
[conda] nvidia-cudnn-cu12 9.5.1.17 pypi_0 pypi
[conda] nvidia-cufft-cu12 11.3.0.4 pypi_0 pypi
[conda] nvidia-curand-cu12 10.3.7.77 pypi_0 pypi
[conda] nvidia-cusolver-cu12 11.7.1.2 pypi_0 pypi
[conda] nvidia-cusparse-cu12 12.5.4.2 pypi_0 pypi
[conda] nvidia-cusparselt-cu12 0.6.3 pypi_0 pypi
[conda] nvidia-nccl-cu12 2.26.2 pypi_0 pypi
[conda] nvidia-nvjitlink-cu12 12.6.85 pypi_0 pypi
[conda] nvidia-nvtx-cu12 12.6.77 pypi_0 pypi
[conda] pytorch-tokenizers 0.1.0 pypi_0 pypi
[conda] torch 2.9.0.dev20250706+cpu pypi_0 pypi
[conda] torchao 0.12.0+gitbc68b11f pypi_0 pypi
[conda] torchaudio 2.8.0.dev20250706+cpu pypi_0 pypi
[conda] torchdata 0.11.0 pypi_0 pypi
[conda] torchsr 1.0.4 pypi_0 pypi
[conda] torchtune 0.6.1 pypi_0 pypi
[conda] torchvision 0.24.0.dev20250706+cpu pypi_0 pypi
[conda] triton 3.3.0 pypi_0 pypi