Skip to content

docs(qnn_backend): update AOT execution flow documentation #628

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
chenghuaWang merged 2 commits into from
Feb 3, 2026
Merged

docs(qnn_backend): update AOT execution flow documentation #628

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
436ec2b
docs(qnn_backend): update AOT execution flow documentation
oreomaker Feb 3, 2026
66adbac
Merge branch 'main' into qnn-aot
chenghuaWang Feb 3, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions docs/qnn_backend/aot_execute.rst
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,8 +18,8 @@ Overall Flow

The QNN AOT execution flow is mainly divided into three stages:

1. **Model Quantization and Export (Python)**: On the host machine, a Python script is used to quantize the pre-trained floating-point model and export it to the MLLM IR (``.mir``) format.
2. **Offline Compilation (C++)**: On the host machine, a C++ compiler program loads the ``.mir`` file, invokes the QNN toolchain for model compilation, graph optimization, and quantization parameter adjustment, and finally generates a QNN Context Binary.
1. **Model Quantization and Export (Python)**: On the host machine, a Python script is used to quantize the pre-trained floating-point model and export it to ``.safetensor`` file. The ``.safetensor`` is then converted to ``.mllm`` file using mllm-convertor.
Copy link
Contributor

@coderabbitai coderabbitai bot Feb 3, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

⚠️ Potential issue | 🟡 Minor

Fix file extension inconsistency.

The file extension should be .safetensors (plural) to match the standard format name and maintain consistency with the rest of the document (lines 33, 42, 44, 49 all use .safetensors).

📝 Proposed fix 
-1.  **Model Quantization and Export (Python)**: On the host machine, a Python script is used to quantize the pre-trained floating-point model and export it to ``.safetensor`` file. The ``.safetensor`` is then converted to ``.mllm`` file using mllm-convertor.
+1.  **Model Quantization and Export (Python)**: On the host machine, a Python script is used to quantize the pre-trained floating-point model and export it to ``.safetensors`` file. The ``.safetensors`` is then converted to ``.mllm`` file using mllm-convertor.
📝 Committable suggestion

‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.

Suggested change
1. **Model Quantization and Export (Python)**: On the host machine, a Python script is used to quantize the pre-trained floating-point model and export it to ``.safetensor`` file. The ``.safetensor`` is then converted to ``.mllm`` file using mllm-convertor.
1. **Model Quantization and Export (Python)**: On the host machine, a Python script is used to quantize the pre-trained floating-point model and export it to ``.safetensors`` file. The ``.safetensors`` is then converted to ``.mllm`` file using mllm-convertor.
🤖 Prompt for AI Agents 
In `@docs/qnn_backend/aot_execute.rst` at line 21, Replace the incorrect singular
extension ".safetensor" with the correct plural ".safetensors" in the sentence
that describes exporting the quantized model (the line mentioning "export it to
``.safetensor`` file"); ensure this matches the other occurrences of
".safetensors" in the document for consistency.

2. **Offline Compilation (C++)**: On the host machine, a C++ compiler program loads the ``.mllm`` file, invokes the QNN toolchain for model compilation, graph optimization, and quantization parameter adjustment, and finally generates a QNN Context Binary.
3. **On-Device Execution (C++)**: On the target device (e.g., a mobile phone), the AOT runner program loads the pre-compiled context binary and executes inference.


Expand Down