Loading_TFCNNViT.py: Python script to load the pre-trained TFCNNViT model and make predictions.
TFCNNViT: Pre-trained model file.
requirements.txt: List of required Python dependencies.
To run the model, install the required Python libraries: The requirements.txt includes: tensorflow-gpu==2.9.0 numpy==1.26.4 pandas==2.2.2 scikit-learn==1.5.2 keras==2.9.0
The model requires the following input features, in this exact order:
time-domain: 3–10 s vertical initial P-wave frequency-domain: The vertical initial P-wave from 3 to 10 seconds is transformed into a power spectrum using NFFT. EpiDist: Epicentral distance (km) Depth: Hypocentral depth (km) Vs30: Site condition (m/s)
Mag.: Magnitude (Mw)
Download the model file:
Place TFCNNViT in a local directory.
Edit the calling script:
Open Loading_TFCNNViT.py in a text editor or IDE (e.g., PyCharm).
Modify the following sections:
data_path: Path to read Dataset.
model_path: Path to TFCNNViT.
save_path: Path to save predict result.
Model configuration:
Ensure TFCNNViT is compatible with the tensorflow version specified in requirements.txt.
For questions or issues, please open an issue on the GitHub repository: https://github.com/lja666/EEWs/TFCNNViT.