fix: handle read-only /data volume, fallback to /tmp/.cache

fix: suppress asyncio fd errors, persistent model cache, increase search_top_k & n_ctx, add linked models

Add sibling/parent directory local fallback paths in RagRetriever

Configure paths to match mounted bucket and add local fallback

Update requirements.txt

Fix HF Space CPU build and Gradio server binding

Optimize UI styling, enable GPU acceleration, native chat completions, and remove thinking panel

Update requirements.txt

Create requirements.txt

Create app.py

initial commit