Cleanup content

Author: odinshen-git

content/learning-paths/cross-platform/multimodel_mnn_v9/1_mnn_v9.md

@@ -1,5 +1,5 @@
 ---
-title: Understand MNN and Omni for Armv9 on-device inference
+title: Multimodal On-Device Inference on Arm v9 with MNN for Audio and Vision
 weight: 2
 
 layout: learningpathall

content/learning-paths/cross-platform/multimodel_mnn_v9/2_mnn_build.md

@@ -14,7 +14,7 @@ mkdir -p ~/mnn_lp
 cd ~/mnn_lp
 ```
 
-Clone MNN and build:
+Clone and build MNN:
 
 ```bash
 git clone https://github.com/alibaba/MNN.git

content/learning-paths/cross-platform/multimodel_mnn_v9/3_mnn_text_baseline.md

@@ -4,7 +4,7 @@ weight: 4
 layout: learningpathall
 ---
 
-## Steps
+## Run the text baseline
 
 Create a simple text prompt file:
 

content/learning-paths/cross-platform/multimodel_mnn_v9/4_mnn_vision.md

@@ -4,7 +4,7 @@ weight: 5
 layout: learningpathall
 ---
 
-## Scenario
+## Prepare a local image asset
 
 In this module you will implement a **smart retail shelf audit** using a single local image.
 
@@ -54,7 +54,7 @@ cat > ~/mnn_lp/prompt_picture_coverage.txt <<'EOF'
 EOF
 ```
 
-Run the vision demo:
+## Run the vision demo:
 
 ```bash
 cd ~/mnn_lp/MNN/build

content/learning-paths/cross-platform/multimodel_mnn_v9/6_mnn_restock_ticket.md

@@ -4,7 +4,7 @@ weight: 7
 layout: learningpathall
 ---
 
-## Goal
+## Create the single-shot multimodal prompt (image + audio)
 
 In this final project you will combine:
 
@@ -44,7 +44,7 @@ cat > ~/mnn_lp/prompt_final_multimodal.txt <<'EOF'
 EOF
 ```
 
-Run the final multimodal demo:
+## Run the final multimodal demo:
 
 ```bash
 cd ~/mnn_lp/MNN/build
@@ -54,14 +54,10 @@ cd ~/mnn_lp/MNN/build
 ./llm_demo ~/mnn_lp/Qwen2.5-Omni-7B-MNN/config.json ~/mnn_lp/prompt_final_multimodal.txt
 ```
 
-This sequence lets you compare:
+## Verify image and audio are both used (optional)
 
-1. Vision-only output  
-2. Audio-only output  
-3. Combined multimodal output  
-
-
-## Verification
+- Swap the **audio** file: quantities/deadline/substitution should change.
+- Swap the **image** file: coverage/priority zone should change.
 
 To verify that the model is truly multimodal, perform a simple A/B test: