Base_Albert/train_results.json

{
    "epoch": 2.0,
    "total_flos": 4371201741330432.0,
    "train_loss": 0.9245093981116634,
    "train_runtime": 4409.6844,
    "train_samples": 131958,
    "train_samples_per_second": 59.849,
    "train_steps_per_second": 0.935
}