Files
g1_timeout_e1_gpt_long_tacc/train_results.json

16 lines
543 B
JSON
Raw Normal View History

{
"achieved_tflops_per_gpu": 0.00380139608872244,
"achieved_tflops_per_gpu_theoretical": 633.5381532479324,
"epoch": 7.0,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.18215620517730713,
"mfu_percent": 0.0002686499002630699,
"mfu_percent_theoretical": 44.77301436381148,
"total_flos": 1416797650550784.0,
"train_loss": 0.3156141554908915,
"train_runtime": 23294.0349,
"train_samples_per_second": 2.983,
"train_steps_per_second": 0.187,
"valid_targets_mean": 6658.9,
"valid_targets_min": 2494
}