qwen2.5-0.5b-anghabench-16kcw-3ep / train_results.json
ahmedheakl's picture
End of training
a1981b0 verified
raw
history blame contribute delete
212 Bytes
{
"epoch": 3.0,
"total_flos": 5.837921389771751e+18,
"train_loss": 0.004886563129512852,
"train_runtime": 102730.9024,
"train_samples_per_second": 14.278,
"train_steps_per_second": 1.785
}