test_results.json

{
    "epoch": 3.0,
    "test_average_metrics": 63.46612466124662,
    "test_em": 54.44444444444444,
    "test_f1": 72.48780487804879,
    "test_loss": 0.19974233210086823,
    "test_runtime": 25.265,
    "test_samples_per_second": 95.943
}