qwen2.5_1.5b_500k_16kcw_2ep / train_results.json
ahmedheakl's picture
End of training
0673e53 verified
raw
history blame contribute delete
210 Bytes
{
"epoch": 2.0,
"total_flos": 1.424923653115478e+19,
"train_loss": 0.003896685916660938,
"train_runtime": 121618.4873,
"train_samples_per_second": 8.04,
"train_steps_per_second": 1.005
}