Files
oh-dcft-v3.1-llama-3.1-nemo…/train_results.json

8 lines
220 B
JSON
Raw Permalink Normal View History

{
"epoch": 2.9993269015032533,
"total_flos": 3503854586953728.0,
"train_loss": 0.5125751896649332,
"train_runtime": 89473.4361,
"train_samples_per_second": 9.562,
"train_steps_per_second": 0.075
}