SymptomsGen is a project that aims to generate medical text related to diseases and their symptoms using a fine-tuned version of the DistilGPT-2 language model. The model is trained on a dataset containing disease names and their corresponding symptoms.
The dataset used in this project is the "Diseases_Symptoms" dataset from the Hugging Face datasets repository. It contains a list of diseases along with their associated symptoms.
The pre-trained language model used in this project is DistilGPT-2, a distilled version of the GPT-2 model. The model is fine-tuned on the "Diseases_Symptoms" dataset using PyTorch and the Hugging Face Transformers library.
To train the model, run the train.py script:
This script will load the dataset, preprocess the data, and train the model for a specified number of epochs. The fine-tuned model will be saved as model.pt.
To generate text using the fine-tuned model, run the inference.py script:
This script will load the fine-tuned model and tokenizer, and generate text based on an input string (e.g., "Kidney Failure").
You can also use the fine-tuned model for inference via a FastAPI apilication. Run the api.py script:
This will start the FastAPI server at http://localhost:8080. You can send a POST request to http://localhost:8080/symptoms with either form data or JSON data to generate text based on the provided input string.
To send form data, make a POST request to http://localhost:8080/generate with the form data containing the input_str field.
To send JSON data, make a POST request to http://localhost:8080/generate with a JSON payload like this:
{
"disease_name": "Kidney Failure"
}
Both requests will return the generated text as a JSON response with the Symptoms key.
- Python 3.6+
- PyTorch
- Hugging Face Transformers
- Datasets
- pandas
- tqdm
- FastAPI
The project uses a config.py file to store the paths to the locally saved tokenizer and pre-trained model files. Create a config.py file in the project directory with the following contents:
TOKENIZER_PATH = 'path/to/tokenizer'
MODEL_PATH = 'path/to/model.pt'Replace 'path/to/tokenizer' and 'path/to/model.pt' with the actual paths to your locally saved tokenizer and pre-trained model files, respectively.
- data.py : Thi sscript is used to preprocess the dataset before training the model.
- train.py: Contains the code for fine-tuning the model.
- inference.py: Contains the code for generating text using the fine-tuned model.
- api.py: Contains the FastAPI apilication for inference.
- config.py: Stores the paths to the locally saved tokenizer and pre-trained model files.
- README.md: This file, providing an overview of the project.
The project includes a Dockerfile that you can use to build a Docker image for the SymptomsGen application. The Dockerfile sets up the necessary environment, installs project dependencies, and configures the FastAPI application to run on port 8080.
To build the Docker image, make sure you have a requirements.txt file listing all the Python dependencies required for your project. Then, run the following command:
docker build -t symptomsgen .
After the image is built, you can run a container from the image with the following command:
docker run -p 8080:8080 symptomsgen
This command will start a container from the symptomsgen image and map the container's port 8080 to the host's port 8080. You should now be able to access your FastAPI application at http://localhost:8080.
The "Diseases_Symptoms" dataset is from the Hugging Face datasets repository. The DistilGPT-2 model is provided by the Hugging Face Transformers library.