执行mllm-convertor --help报错，pymllm报错

[singlecatlmx]

安装QNN和Hexagon NPU SDK，设置了环境变量。
bash ./scripts/install_pymllm.sh安装pymllm后，执行mllm-convertor --help报错
Traceback (most recent call last):
File "/home/lmx/Softwares/anaconda3/envs/qnn/bin/mllm-convertor", line 3, in
from pymllm.utils.mllm_convertor import main
File "/home/lmx/Softwares/anaconda3/envs/qnn/lib/python3.10/site-packages/pymllm/init.py", line 6, in
from . import ffi
File "/home/lmx/Softwares/anaconda3/envs/qnn/lib/python3.10/site-packages/pymllm/ffi/init.py", line 324, in
class QnnDeviceAndContext(tvm_ffi.Object):
File "/home/lmx/Softwares/anaconda3/envs/qnn/lib/python3.10/site-packages/tvm_ffi/registry.py", line 56, in register
raise ValueError(f"Cannot find object type index for {object_name}")
ValueError: Cannot find object type index for mllm.qualcomm.QnnDeviceAndContext

[coderabbitai]

📝 CodeRabbit Plan Mode

Generate an implementation plan and prompts that you can use with your favorite coding agent.

• Create Plan

Examples

• Example 1
• Example 2

_{You can configure auto-planning by selecting labels in the issue_enrichment configuration.}

🔗 Similar Issues

Related Issues

• 自己转换的qwen2.5输出结果是乱码 #456
• 1+12 8gen3 Failed to create device #341
• Android 8gen3 NPU上推理卡死 #309
• “Failed to register op package” Error when running./demo_qwen_npu createContext() #274
• 视觉权重无法找到，请问 demo_qwen2_vl_npu.cpp 中的模型文件如何获取 #535
• QNN模型初始化失败 #272

👤 Suggested Assignees

• zhenchong
• 985958118
• finneyyan
• HirokenOvo
• 970814

---

🧪 Issue enrichment is currently in open beta.

To disable automatic issue enrichment, add the following to your .coderabbit.yaml:

issue_enrichment:
  auto_enrich:
    enabled: false

💬 Have feedback or questions? Drop into our discord or schedule a call!

[chenghuaWang]

Hello singlecatlmx:

This issue was introduced in the Qnn AOT PR. The main reason is that the __init__.py file in pymllm does not consider whether the mllm x86-qnn-aot extension has been compiled. I am currently fixing this bug.

[chenghuaWang]

fixed in #563 @singlecatlmx