Skip to content

Basic RNN

Basic RNN model

Basic RNN for the next Word Prediction

Model Training

  • Train the model using below command:
python model.py
  • Model gets saved as keras format.

Evaluate Model

  • Test the model using below command:
python evaluate.py

Visual Explaination

Training

flowchart TD
    A[Start Script] --> B[Prepare Text Corpus]
    B --> C[Tokenize Corpus with Keras Tokenizer]
    C --> D[Generate Input Sequences and Targets]
    D --> E[Pad Sequences to Fixed Length]
    E --> F[Convert Sequences to Numpy Arrays]

    F --> G[Build RNN Model with Embedding + SimpleRNN + Dense]
    G --> H[Compile Model with Adam and Crossentropy Loss]
    H --> I[Train Model on Input Data]
    I --> J[Save Trained Model as .keras]

Testing

flowchart TD
    A[Start Script] --> B[Load Trained RNN Model from File]
    B --> C[Recreate Corpus and Tokenizer]
    C --> D[Reconstruct Index-to-Word Mapping]
    D --> E[User Enters Input Phrase]
    E --> F[Tokenize and Pad the Input Phrase]
    F --> G[Predict Next Word Using Model]
    G --> H[Find Predicted Word from Index]
    H --> I[Display Predicted Word to User]
    I --> E