DUAL-GPO
/

phi-2-gpo-test-longest-iter-v1-4

alignment-handbook

Generated from Trainer

Model card Files Files and versions Community

phi-2-gpo-test-longest-iter-v1-4 / train_results.json

BraylonDash's picture

Model save

50e9903 verified 3 months ago

raw history blame contribute delete

No virus

196 Bytes

	{
	"epoch": 3.97,
	"train_loss": 0.001002159876318934,
	"train_runtime": 2735.225,
	"train_samples": 61135,
	"train_samples_per_second": 1.462,
	"train_steps_per_second": 0.091
	}