training-complete.json

{
  "completed_steps": 500,
  "final_loss": 0.11363611370325089,
  "final_teacher_loss": 0.05577564239501953,
  "final_ground_truth_loss": 0.28721752762794495
}