Files
a1-swesmith/train_results.json
ModelHub XC d48422608e 初始化项目,由ModelHub XC社区提供模型
Model: DCAgent/a1-swesmith
Source: Original Platform
2026-05-07 09:21:00 +08:00

16 lines
544 B
JSON

{
"achieved_tflops_per_gpu": 0.004657515168383943,
"achieved_tflops_per_gpu_theoretical": 687.7446262531031,
"epoch": 7.0,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.14940977096557617,
"mfu_percent": 0.0003291530154334942,
"mfu_percent_theoretical": 48.603860512586785,
"total_flos": 1139548609970176.0,
"train_loss": 0.2251253145215901,
"train_runtime": 15291.7995,
"train_samples_per_second": 3.246,
"train_steps_per_second": 0.203,
"valid_targets_mean": 3384.2,
"valid_targets_min": 497
}