This repo contains code for a toxicity grading model using TensorFlow and Gradio. The model is trained to detect whether a comment is toxic, obscene, threatening, or involves identity hate. The steps below outline the process of installing dependencies, preprocessing the data, creating a sequential model, making predictions, evaluating the model, and testing it using Gradio.
Before running the code, please make sure to install the required dependencies by executing the following command:
!pip install tensorflow tensorflow-gpu pandas matplotlib sklearn
The preprocessing step involves converting the comment text into numerical features using the TextVectorization layer from TensorFlow. The data is then split into training, validation, and test datasets.
The sequential model is defined using the TensorFlow Keras API. It consists of an embedding layer, a bidirectional LSTM layer, fully connected layers for feature extraction, and a final output layer with sigmoid activation. The model is compiled with binary cross-entropy loss and the Adam optimizer.
The model is trained using the training and validation datasets. The training progress is visualized using a line plot showing the loss and metrics over epochs.
The model can be used to make predictions on new comment data. The code provides examples of how to input a single comment or a batch of comments to obtain toxicity predictions.
The model's performance is evaluated using precision, recall, and accuracy metrics on the test dataset.
The code installs the Gradio library and saves the trained model. It then loads the model and defines a function to score comments based on toxicity. Gradio is used to create a user-friendly interface where users can input a comment and see the toxicity grades for different categories.
To run the Gradio interface, please make sure to install the required dependencies by executing the command:
!pip install gradio jinja2
Once the interface is launched, users can input comments and obtain the toxicity grades in different categories.