Skip to content

Error occurred when running Qwen 3.5 examples #18832

@Mingle29

Description

@Mingle29

Excuse me. I'm trying to run the example of Qwen 3.5. I managed to export the .pte model. But when I try to run the native Runner:

python -m executorch.examples.models.llama.runner.native \
  --model qwen3_5_0_8b \
  --pte qwen3_5_0_8b_fp32.pte \
  --tokenizer /path/to/tokenizer.json \
  --tokenizer_config /path/to/tokenizer_config.json \
  --prompt "<|im_start|>user\nHello<|im_end|>\n<|im_start|>assistant\n" \
  --params examples/models/qwen3_5/config/0_8b_config.json \
  --max_len 128 \
  -kv \
  --temperature 0.3

I met the error below:

Traceback (most recent call last):
File "/.../executorch/extension/llm/custom_ops/custom_ops.py", line 26, in
op = torch.ops.llama.sdpa_with_kv_cache.default
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/.../miniforge3/envs/execut_main_new/lib/python3.12/site-packages/torch/_ops.py", line 1379, in getattr
raise AttributeError(
AttributeError: '_OpNamespace' 'llama' object has no attribute 'sdpa_with_kv_cache'
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "", line 198, in _run_module_as_main
File "", line 88, in _run_code
File "/.../executorch/examples/models/llama/runner/native.py", line 26, in
from executorch.extension.llm.custom_ops import custom_ops # noqa # usort: skip
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/.../executorch/extension/llm/custom_ops/custom_ops.py", line 44, in
assert len(libs) == 1, f"Expected 1 library but got {len(libs)}"
^^^^^^^^^^^^^^
AssertionError: Expected 1 library but got 0

Metadata

Metadata

Labels

No labels
No labels

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions