search

synlp / R2-LLM

The official GitHub repository of the AAAI-2024 paper "Bootstrapping Large Language Models for Radiology Report Generation".
MIT License
41 stars 2 forks source link

questions for the reproduction #5

Open Andyfever123 opened 5 months ago

Andyfever123 commented 5 months ago

your paper ''Bootstrapping Large Language Models for Radiology Report Generation''is so excellent that I want to reproduce your code from scratch, but I am confused about some of the configs, such as the config files for the iu-xray dataset and mimic dataset"align.yaml" and "generate_then_refine.yaml". About this "
build_info: storage: /path/to/mimic unlabeled_annotation_path: /path/to/pubmed" could you please tell me what is the type of the pubmed?Only reports or both images and reports? And how can I download the pubmed you used?Also, config.yaml for stage1 and stage2 in train_configs, regarding "ckpt: '/path/to/linear/layer'", how should I fill in this ckpt path correctly?

strawberrylili commented 2 months ago

your paper ''Bootstrapping Large Language Models for Radiology Report Generation''is so excellent that I want to reproduce your code from scratch, but I am confused about some of the configs, such as the config files for the iu-xray dataset and mimic dataset"align.yaml" and "generate_then_refine.yaml". About this " build_info: storage: /path/to/mimic unlabeled_annotation_path: /path/to/pubmed" could you please tell me what is the type of the pubmed?Only reports or both images and reports? And how can I download the pubmed you used?Also, config.yaml for stage1 and stage2 in train_configs, regarding "ckpt: '/path/to/linear/layer'", how should I fill in this ckpt path correctly?

Hi, I also meet the same problem. Have you solved it now? If it is possible, shall we have a small discussion about that? thx a lot!