Files
b1_top8_seq/train_results.json
ModelHub XC ce6173d174 初始化项目,由ModelHub XC社区提供模型
Model: DCAgent/b1_top8_seq
Source: Original Platform
2026-04-23 20:40:07 +08:00

16 lines
545 B
JSON

{
"achieved_tflops_per_gpu": 0.003671455564786977,
"achieved_tflops_per_gpu_theoretical": 623.3744395594629,
"epoch": 7.0,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.17723463475704193,
"mfu_percent": 0.0002594668243665708,
"mfu_percent_theoretical": 44.054730710916104,
"total_flos": 1286594594209792.0,
"train_loss": 0.3047364686883014,
"train_runtime": 21901.9843,
"train_samples_per_second": 2.937,
"train_steps_per_second": 0.184,
"valid_targets_mean": 3156.8,
"valid_targets_min": 1012
}