phi-ft-1000000-fp-newsplit / train_results.json
KaranChand's picture
Model save
8591e44 verified
{
"epoch": 1.0,
"total_flos": 1.6067287852253184e+18,
"train_loss": 0.5878726873860691,
"train_runtime": 61938.6999,
"train_samples_per_second": 0.564,
"train_steps_per_second": 0.141
}