Files
a1-nemotron_csharp/train_results.json

16 lines
545 B
JSON
Raw Normal View History

{
"achieved_tflops_per_gpu": 0.0030934865769747305,
"achieved_tflops_per_gpu_theoretical": 552.2284437401764,
"epoch": 7.0,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.2052934765815735,
"mfu_percent": 0.00021862095950351453,
"mfu_percent_theoretical": 39.026745140648515,
"total_flos": 1298839445962752.0,
"train_loss": 0.2040717954099597,
"train_runtime": 26241.4151,
"train_samples_per_second": 2.601,
"train_steps_per_second": 0.163,
"valid_targets_mean": 4578.3,
"valid_targets_min": 889
}