d1_science_shortest_3k / train_results.json
ryanmarten's picture
Upload model
fb148d5 verified
{
"epoch": 7.0,
"total_flos": 2.93440642585985e+17,
"train_loss": 0.36575089378919434,
"train_runtime": 6464.1361,
"train_samples_per_second": 3.422,
"train_steps_per_second": 0.036
}