unraveled-7b-dpo-lora / train_results.json
Ber Zoidberg
Model save
b445297
raw
history blame contribute delete
194 Bytes
{
"epoch": 3.0,
"train_loss": 0.622471270236102,
"train_runtime": 20371.8366,
"train_samples": 61966,
"train_samples_per_second": 9.125,
"train_steps_per_second": 0.036
}