titulm-llama-3.2-3b-v1.1 / train_results.json
SKNahin's picture
End of training
f88b157 verified
raw
history blame contribute delete
220 Bytes
{
"epoch": 0.9999197238500441,
"total_flos": 9004996003627008.0,
"train_loss": 0.728125217524414,
"train_runtime": 85870.4917,
"train_samples_per_second": 24.371,
"train_steps_per_second": 0.054
}