qwen2.5_1.5b_500k_16kcw_4ep / train_results.json
ahmedheakl's picture
End of training
9695937 verified
raw
history blame contribute delete
210 Bytes
{
"epoch": 4.0,
"total_flos": 2.849849363030396e+19,
"train_loss": 0.00272644101527126,
"train_runtime": 243116.2392,
"train_samples_per_second": 8.044,
"train_steps_per_second": 1.006
}