Error in pip editable mode in export_llama

[iseeyuan]

🐛 Describe the bug

Repro: have a fresh ET git clone.

git clone https://github.com/pytorch/executorch.git
cd executorch

git submodule sync
git submodule update --init

./install_executorch.sh --editable

Run the command below and got the error:

python -m examples.models.llama.export_llama -p /Users/myuan/data/stories_110M/params.json -c /Users/myuan/data/stories_110M/stories110M.pt -X --xnnpack-extended-ops -qmode 8da4w -G 128 --use_kv_cache --use_sdpa_with_kv_cache --verbose --output_name test_sdpa_with_kv.pte 
Traceback (most recent call last):
  File "/Users/myuan/src/executorch/exir/dialects/_ops.py", line 100, in __getattr__
    parent_packet = getattr(self._op_namespace, op_name)
  File "/Users/myuan/miniconda3/envs/executorch/lib/python3.10/site-packages/torch/_ops.py", line 1267, in __getattr__
    raise AttributeError(
AttributeError: '_OpNamespace' 'quantized_decomposed' object has no attribute 'quantize_per_channel'

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "/Users/myuan/miniconda3/envs/executorch/lib/python3.10/runpy.py", line 196, in _run_module_as_main
    return _run_code(code, main_globals, None,
  File "/Users/myuan/miniconda3/envs/executorch/lib/python3.10/runpy.py", line 86, in _run_code
    exec(code, run_globals)
  File "/Users/myuan/src/executorch/examples/models/llama/export_llama.py", line 20, in <module>
    from .export_llama_lib import build_args_parser, export_llama

The above command works well without the --editable option.
File "/Users/myuan/src/executorch/examples/models/llama/export_llama_lib.py", line 25, in
from executorch.backends.vulkan._passes.remove_asserts import remove_asserts
File "/Users/myuan/src/executorch/backends/vulkan/init.py", line 7, in
from .partitioner.vulkan_partitioner import VulkanPartitioner
File "/Users/myuan/src/executorch/backends/vulkan/partitioner/vulkan_partitioner.py", line 16, in
from executorch.backends.vulkan.op_registry import (
File "/Users/myuan/src/executorch/backends/vulkan/op_registry.py", line 225, in
exir_ops.edge.quantized_decomposed.quantize_per_channel.default,
File "/Users/myuan/src/executorch/exir/dialects/_ops.py", line 104, in getattr
raise AttributeError(
AttributeError: '_OpNamespace' 'edge.quantized_decomposed' object has no attribute 'quantize_per_channel'
ERROR conda.cli.main_run:execute(47): conda run python -m examples.models.llama.export_llama -p /Users/myuan/data/stories_110M/params.json -c /Users/myuan/data/stories_110M/stories110M.pt -X --xnnpack-extended-ops -qmode 8da4w -G 128 --use_kv_cache --use_sdpa_with_kv_cache --verbose --output_name test_sdpa_with_kv.pte failed. (See above for error)

Process finished with exit code 1

If I do

Versions

Collecting environment information...
PyTorch version: 2.7.0.dev20250311
Is debug build: False
CUDA used to build PyTorch: None
ROCM used to build PyTorch: N/A

OS: macOS 15.3.2 (arm64)
GCC version: Could not collect
Clang version: 16.0.0 (clang-1600.0.26.3)
CMake version: version 3.31.4
Libc version: N/A

Python version: 3.10.16 (main, Dec 11 2024, 10:22:29) [Clang 14.0.6 ] (64-bit runtime)
Python platform: macOS-15.3.2-arm64-arm-64bit
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
HIP runtime version: N/A
MIOpen runtime version: N/A
Is XNNPACK available: True

CPU:
Apple M1 Max

Versions of relevant libraries:
[pip3] executorch==0.6.0a0+9a0c2db
[pip3] flake8==6.1.0
[pip3] flake8-breakpoint==1.1.0
[pip3] flake8-bugbear==24.4.26
[pip3] flake8-comprehensions==3.14.0
[pip3] flake8-plugin-utils==1.3.3
[pip3] flake8-pyi==23.5.0
[pip3] mypy==1.14.1
[pip3] mypy-extensions==1.0.0
[pip3] numpy==2.2.2
[pip3] torch==2.7.0.dev20250311
[pip3] torchao==0.10.0+git7d879462
[pip3] torchaudio==2.6.0.dev20250311
[pip3] torchsr==1.0.4
[pip3] torchtune==0.5.0
[pip3] torchvision==0.22.0.dev20250311
[conda] executorch 0.6.0a0+9a0c2db pypi_0 pypi
[conda] numpy 2.2.2 pypi_0 pypi
[conda] torch 2.7.0.dev20250311 pypi_0 pypi
[conda] torchao 0.10.0+git7d879462 pypi_0 pypi
[conda] torchaudio 2.6.0.dev20250311 pypi_0 pypi
[conda] torchfix 0.6.0 pypi_0 pypi
[conda] torchsr 1.0.4 pypi_0 pypi
[conda] torchtune 0.5.0 pypi_0 pypi
[conda] torchvision 0.22.0.dev20250311 pypi_0 pypi

cc @larryliu0820 @jathu @lucylq

[larryliu0820]

I think the issue is at the relative import from .export_llama_lib import build_args_parser, export_llama. I'll try to see if it's fixed after changing it to full import.

[larryliu0820]

Unfortunately this is the same issue as #9558. We need to make progress on #8699 in order to make this work.

[larryliu0820]

This is mitigated by using symlink, introduced in #9818. Closing