Fadil369/medical-instruction-100k

--- license: mit language: - en tags: - medi - medical pretty_name: python size_categories: - 10K<n<100K --- # What is the Dataset About?🤷🏼‍♂️ --- The dataset is useful for training a Generative Language Model for the Medical application and instruction purposes, the dataset consists of various thoughs proposed by the people [**mentioned as the Human** ] and there responses including Medical Terminologies not limited to but including names of the drugs, prescriptions, yogic exercise suggessions, breathing exercise suggessions and few natural home made prescriptions. # How the Dataset was made?😅 --- I have used all the available opensource datasets and combined them into a single datsource for training, which is completely opensourced and somewhat reliable. * There is another refined and updated version of this datset here 👉🏼 [Link](https://huggingface.co/datasets/Mohammed-Altaf/medical-instruction-120k) ## Example Training Scripts: * Qlora Fine Tuning - ## Tips: This is my first dataset to upload on HuggingFace, so below are the thing I wish I could have known * always save your final dataset before uploading to hub as a json with lines. * The json should have the records orientation, which will be helpful while loading the dataset properly without any error. ```{python} # use below if you are using pandas for data manipulation train.to_json("dataset_name.json", orient='records', lines=True) test.to_json("dataset_name.json", orient='records', lines=True) ```