📁 BP-GPT Implementation

⚠️ File Notes

• The GP2model folder exceeds GitHub's size limit and cannot be uploaded directly.
• Missing .bin file (pytorch_model.bin) can be downloaded from:
  Hugging Face GPT-2 Model

🛠️ Requirements

• Python 3.8
• PyTorch 2.4.1 (CUDA 12.1)

🚀 Workflow

1. Data Generation

Run main_hybrid.m to generate:

• Training dataset
• Test dataset

2. Model Training

Execute LLM4BF.py to train the BP-GPT model.

3. Model Testing

Run test.py to evaluate predicted beam performance.

📊 Performance Analysis

Temporal Performance

• TindexRate.m: Generates Spectral Efficiency (SE) vs. time steps.

Velocity Impact

• Velocity.m: Plots SE vs. different velocities.

Noise Robustness

• PNRRate.m: Calculates average SE across different test SNRs.