qwen2.5_0.5b_500k_O3_16kcw_2ep_2 / train_results.json
ahmedheakl's picture
End of training
48050b7 verified
raw
history blame contribute delete
223 Bytes
{
"epoch": 1.9999915560152837,
"total_flos": 3.430800695853318e+18,
"train_loss": 0.01405828405903957,
"train_runtime": 96268.9774,
"train_samples_per_second": 9.841,
"train_steps_per_second": 2.46
}