Quality Control Codes for Medical Imaging Report Text Standards

The overall workflow of the code is as follows: first, run teacher-model inference.cy to complete the teacher model inference; Then execute POST - inference processing. py to process the output results of the teacher model; Then run train.sh to train the student model; Afterwards, start infra_student_madel-vllm.py to carry out student model inference. Evaluate.py is used for model evaluation. The prompts for synthetic data generation.doc file stores the prompt words used for generating synthetic data.

Bootstrap. py displays the calculation method for confidence intervals. The conventional all rules prompting and rule decomposed LLM only prompting folder stores the prompt words for ablation experiments.