deploy: put HF Spaces YAML front-matter at top of README

deploy: drop personal notes from public Space

Week4: Gradio 前端 + PDF 入库 + HF Spaces 部署件

Week3: FastAPI layer + wire RAG into graph qa node

Week3: LangGraph parallel orchestration + per-agent model routing

Add before/after eval: serve LoRA adapter + base vs finetune comparison

Add Week2 agent + RAG layers; record training teardown

Disable periodic eval during training

Make training resumable + tune for RTX4090 24G spot instance

Add LLaMA-Factory training config for Qwen2.5-7B LoRA

Fix data leakage and label consistency before training

Add preprocessing script: ad removal, label normalization, dedup

Add WORKFLOW.md: architecture & pipeline diagrams with tool selection rationale

Add dataset download script; switch data source to FinCUGE+BAAI

Update progress log: local Qwen2.5-7B baseline test passed

Add FinSight project plan