gemma7b-summarize-gpt4o-8k / train_results.json
juyongjiang's picture
update model checkpoint
dcd490a verified
raw
history blame
234 Bytes
{
"epoch": 10.0,
"total_flos": 4.268849030789857e+17,
"train_loss": 7.876109651156834,
"train_runtime": 340.0833,
"train_samples": 8076,
"train_samples_per_second": 25.758,
"train_steps_per_second": 0.412
}