{"id":25416677,"url":"https://github.com/HeathSun/MediQA-Model","last_synced_at":"2025-10-31T09:30:49.834Z","repository":{"id":38322536,"uuid":"506526859","full_name":"HShawnSun/MediQA-Model","owner":"HShawnSun","description":"2022 May-Aug, NLP Summer Research Assistant at CS \u0026 AI Lab, UNNC. Explored the feasibility of Bert-like models for machine reading comprehension of small  text in medical areas.","archived":false,"fork":false,"pushed_at":"2025-01-16T03:41:14.000Z","size":104,"stargazers_count":3,"open_issues_count":0,"forks_count":0,"subscribers_count":2,"default_branch":"gh-pages","last_synced_at":"2025-01-16T04:36:10.214Z","etag":null,"topics":["deep-learning","natural-language-processing","pytorch","tensorflow"],"latest_commit_sha":null,"homepage":"","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/HShawnSun.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null}},"created_at":"2022-06-23T06:44:17.000Z","updated_at":"2025-01-16T03:41:16.000Z","dependencies_parsed_at":"2022-08-25T02:40:20.379Z","dependency_job_id":null,"html_url":"https://github.com/HShawnSun/MediQA-Model","commit_stats":null,"previous_names":["hshawnsun/mediqa-model"],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/HShawnSun%2FMediQA-Model","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/HShawnSun%2FMediQA-Model/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/HShawnSun%2FMediQA-Model/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/HShawnSun%2FMediQA-Model/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/HShawnSun","download_url":"https://codeload.github.com/HShawnSun/MediQA-Model/tar.gz/refs/heads/gh-pages","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":239163390,"owners_count":19592344,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["deep-learning","natural-language-processing","pytorch","tensorflow"],"created_at":"2025-02-16T16:56:07.943Z","updated_at":"2025-10-31T09:30:44.521Z","avatar_url":"https://github.com/HShawnSun.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"\r\n\r\n# **Bert-like Models Feasibility Exploration on Medical Short-Text Q\u0026A Tasks**\r\n\r\nIn this project, I explored the feasibility of leveraging Bert-like models for machine reading comprehension (MRC) in specialized domains, specifically medical short-text Q\u0026A tasks. The work focused on blending different classes of Bert-like models to improve performance when handling small medical datasets extracted from medical illustration videos.\r\n\r\nKey accomplishments and findings include:\r\n\r\n- **Model Exploration:** Selected and blended various Bert-like models to enhance performance on small text data from medical contexts.\r\n- **Field-Specific MRC Model Training:** Attempted to train a specialized MRC model tailored to medical Q\u0026A tasks, adjusting hyperparameters and evaluating performance.\r\n- **Performance Testing:** Rigorous testing and evaluation of the model’s ability to comprehend and respond to medical questions.\r\n- **Skill Development:** Improved my Natural Language Processing (NLP) capabilities by working closely with my supervisor and Ph.D. students. Gained a deeper understanding of deep learning (DL) and reinforcement learning (RL) formulas.\r\n- **Research Contribution:** This research contributed to the Ningbo 2025 Science and Technology Innovation Program, advancing knowledge in the application of Bert-like models to medical tasks.\r\n\r\n---\r\n\r\n### **Prediction Steps**\r\n\r\nTo execute predictions with the trained model, follow these steps:\r\n\r\n1. **Run the Prediction Scripts:**\r\n   - For **关黄母颗粒** question-answering prediction:\r\n     ```bash\r\n     sh test_bert_ghm.sh\r\n     ```\r\n   - For **丁苯酚** question-answering prediction:\r\n     ```bash\r\n     sh test_bert_dbf.sh\r\n     ```\r\n\r\n2. **Adjust Storage Path**: When executing the script, make sure to change the storage address as prompted by `metrics.py` at line 730.\r\n\r\n---\r\n\r\n### **Parameters Overview**\r\n\r\n- **`--lm`**: Specify the folder containing the pre-trained model to be loaded.\r\n- **`--do_train`**: Activate training mode.\r\n- **`--evaluate_during_training`**: Enable validation during training.\r\n- **`--do_test`**: Activate prediction mode.\r\n- **`--version_2_with_negative`**: Adapt the model for datasets that may contain both questions with answers and questions without answers.\r\n- **`--threads`**: Specify the number of threads for data processing.\r\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2FHeathSun%2FMediQA-Model","html_url":"https://awesome.ecosyste.ms/projects/github.com%2FHeathSun%2FMediQA-Model","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2FHeathSun%2FMediQA-Model/lists"}