test_results.json

{
    "epoch": 3.0,
    "test_average_metrics": 75.00792291086177,
    "test_em": 74.50579862941487,
    "test_f1": 75.51004719230866,
    "test_loss": 0.48300713300704956,
    "test_runtime": 342.8147,
    "test_samples_per_second": 44.269
}