64k_0.5B / train_results.json
Albus-Chen's picture
Initial upload of trained model
cb3380b verified
{
"epoch": 3.0,
"total_flos": 4.3556612022154035e+18,
"train_loss": 0.6027277402877808,
"train_runtime": 90142.7719,
"train_samples_per_second": 2.13,
"train_steps_per_second": 0.033
}