pythia-6.9b-HC3 / train_results.json
Peter Szemraj
add sharded checkpoint
4b90790
raw
history blame
195 Bytes
{
"epoch": 1.99,
"train_loss": 1.1195918215981013,
"train_runtime": 35849.6176,
"train_samples": 5097,
"train_samples_per_second": 0.284,
"train_steps_per_second": 0.004
}