Revert to HF free embedding (sentence-transformers) for Weaviate retrieval

Add PNG files as regular binary files (not LFS)

Add Weaviate index rebuild instructions

Switch to OpenAI embedding (text-embedding-3-small) for Weaviate retrieval

Switch to React product UI: server.py, web/, Dockerfile, FastAPI deps (PNG as normal blobs)

fix: 修复 history 格式转换边界情况，添加防御性检查

fix: Gradio 6.0 messages 格式兼容 - 添加 tuples/messages 转换

fix: Gradio 6.0 兼容性 + 安装 libxcb 系统库

fix: 使用 Docker 构建，强制 huggingface_hub>=1.3.0

fix: 强制 huggingface_hub>=1.3.0 满足 transformers 要求

fix: 兼容旧版 huggingface_hub，注入 is_offline_mode 解决 sentence-transformers ImportError

fix: 升级 huggingface_hub>=1.2.0 解决 is_offline_mode ImportError

fix: Weaviate 检索无结果 - 预热 embedding、超时 45s、改进错误提示

fix: Weaviate 连接优化 - 缓存 embedding、超时、UI 状态显示

feat: ClareVoice 直连 Weaviate Cloud，无需 GenAICoursesDB Space

Improve references precision and avoid irrelevant RAG citations

GenAICoursesDB 调用增加 25 秒超时，避免冷启动阻塞

集成 GenAICoursesDB 向量知识库（方案三）

Fix podcast regex error and optimize embedding generation

Add vector database (FAISS) and enhanced PDF parsing (unstructured.io)

Add TTS and podcast generation

Update app.py

Update app.py

Update app.py

Update requirements.txt

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update requirements.txt

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update rag_engine.py

Update app.py

Update app.py

Update rag_engine.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py