chessgpt2-medium-l / train_results.json
dakwi's picture
End of training
350a17b verified
raw
history blame contribute delete
241 Bytes
{
"epoch": 3.0,
"total_flos": 1.8462142604897157e+18,
"train_loss": 0.19927044710540576,
"train_runtime": 48883.7053,
"train_samples": 1000192,
"train_samples_per_second": 61.382,
"train_steps_per_second": 0.959
}