chessgpt-medium-l / train_results.json
dakwi's picture
End of training
ef644c2 verified
raw
history blame contribute delete
235 Bytes
{
"epoch": 1.0,
"total_flos": 6.467778978579087e+17,
"train_loss": 0.0316199453125,
"train_runtime": 3118.7762,
"train_samples": 1000000,
"train_samples_per_second": 320.639,
"train_steps_per_second": 5.01
}