training_meta.json

{
  "base_model": "tiiuae/Falcon-H1-0.5B-Base",
  "method": "DPO",
  "dataset_size": 4234,
  "epochs": 3,
  "lora_r": 16,
  "lora_alpha": 32,
  "beta": 0.1,
  "final_loss": 0.03056402061143363
}