Model: open-sci/sft__ot30k_Qwen2.5-1.5B-DPO-Tulu3-decontaminated Source: Original Platform
12 lines
338 B
JSON
12 lines
338 B
JSON
{
|
|
"epoch": 5.0,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.48496419191360474,
|
|
"total_flos": 1940656491724800.0,
|
|
"train_loss": 1.0710605895772893,
|
|
"train_runtime": 5191.8253,
|
|
"train_samples_per_second": 28.892,
|
|
"train_steps_per_second": 0.226,
|
|
"valid_targets_mean": 11353.3,
|
|
"valid_targets_min": 2084
|
|
} |