qwen2.5-0.5b-expo-DPO-ES-0.01 / all_results.json
hZzy's picture
End of training
0fb2aea verified
{
"epoch": 1.5588096362777515,
"eval_dpo_loss": 0.6902037858963013,
"eval_logits": -1.7611440420150757,
"eval_logps": -99.38214111328125,
"eval_loss": 0.6893583536148071,
"eval_objective": 0.6902037858963013,
"eval_ranking_idealized": 0.8731883764266968,
"eval_ranking_idealized_expo": 0.5320910811424255,
"eval_ranking_simple": 0.5305383205413818,
"eval_regularize": 0.6902037858963013,
"eval_runtime": 309.3664,
"eval_samples": 5790,
"eval_samples_per_second": 18.716,
"eval_steps_per_second": 1.561,
"eval_wo_beta": 9.357504844665527,
"total_flos": 0.0,
"train_loss": 0.17483963706276634,
"train_runtime": 6877.8936,
"train_samples": 50802,
"train_samples_per_second": 36.931,
"train_steps_per_second": 0.256
}