refactor: update eval_baseline.py to integrate Ollama for local LLM calls and improve usage instructions

add baseline eval script and results