Update README with QALoop architecture diagram via GitHub URL

HF Space rejects binary git pushes; reference figure from GitHub monorepo instead.

Co-authored-by: Cursor <cursoragent@cursor.com>

.gitignore

@@ -22,5 +22,8 @@ data/
 *.db
 seed/*.db
 
+# 图片（HF Space 不接受二进制 git push，保留在 GitHub monorepo）
+seed/*.png
+
 # 环境变量
 .env

README.md

@@ -128,6 +128,10 @@ client = OpenAI(
 
 ## System Overview
 
+![QALoop System Architecture](https://raw.githubusercontent.com/JackKuo666/QALoop/main/platform/seed/figure1.png)
+
+The diagram above shows the full QALoop workflow. This platform corresponds to the **Expert Validation Platform** stage — receiving candidate QA pairs from upstream synthesis pipelines and supporting human quality review, feedback, and export.
+
 This repository covers **data import → schema configuration → multi-user annotation → statistics/analysis → export**. It does not include built-in "automatic QA generation" or "automatic model scoring/evaluation" modules; these can be integrated via upstream/downstream systems.
 
 ```mermaid

README_zh.md

@@ -128,6 +128,10 @@ client = OpenAI(
 
 ## 系统概览
 
+![QALoop 系统架构](https://raw.githubusercontent.com/JackKuo666/QALoop/main/platform/seed/figure1.png)
+
+上图展示了 QALoop 完整流程。本平台对应其中的 **专家验证平台（Expert Validation Platform）** 环节——接收上游 QA 合成管道产出的候选 QA 对，支持人工质检、反馈与导出。
+
 本仓库覆盖 **数据导入 → 配置量表 → 多人标注 → 统计/分析 → 导出**；不包含内置的「自动 QA 生成」或「模型自动打分评测」模块，二者可通过上游/下游系统接入。
 
 ```mermaid