a1_science_fineweb / train_results.json
ryanmarten's picture
End of training
bfa9729 verified
raw
history blame contribute delete
224 Bytes
{
"epoch": 4.988354430379747,
"total_flos": 1.2467177959037338e+18,
"train_loss": 0.33969098806865816,
"train_runtime": 44982.8499,
"train_samples_per_second": 3.512,
"train_steps_per_second": 0.027
}