InstruBPM enables BPMN model generation via instruction tuning

InstruBPM uses parameter-efficient fine-tuning and quantization to enable open-weight LLMs to generate BPMN process models from natural language. The approach allows for high-fidelity, privacy-preserving, on-premise deployment for software engineering workflows.