llama-8b-dpo-full / all_results.json
jkazdan's picture
Model save
0de381b verified
{
"epoch": 0.9936305732484076,
"total_flos": 0.0,
"train_loss": 0.386262208987505,
"train_runtime": 654.7511,
"train_samples": 9999,
"train_samples_per_second": 15.271,
"train_steps_per_second": 0.119
}