d1_science_gpt_10k / train_results.json
ryanmarten's picture
Upload model
4dd8ec5 verified
{
"epoch": 4.992,
"total_flos": 6.781150607693578e+17,
"train_loss": 0.38343327595637394,
"train_runtime": 12509.8929,
"train_samples_per_second": 3.997,
"train_steps_per_second": 0.031
}