A Flexible Framework for Experiencing Heterogeneous LLM Inference/Fine-tune Optimizations - View it on GitHub