baseline-Llama-3-8B-Instruct-sft / train_results.json
ZhangShenao's picture
Model save
82fc17e verified
raw
history blame contribute delete
255 Bytes
{
"epoch": 2.9977761304670127,
"total_flos": 1.1934072323532915e+19,
"train_loss": 0.20480146514054692,
"train_runtime": 12349.0973,
"train_samples": 395000,
"train_samples_per_second": 10.482,
"train_steps_per_second": 0.082
}