qwen2-math-1_5b-step-dpo / all_results.json
rasdani's picture
Model save
40e5cfe verified
raw
history blame contribute delete
232 Bytes
{
"epoch": 7.964444444444444,
"total_flos": 0.0,
"train_loss": 0.19260043763954723,
"train_runtime": 3828.823,
"train_samples": 10795,
"train_samples_per_second": 22.555,
"train_steps_per_second": 0.351
}