ondevicellm
/

tinyllama_mole_dpo_ep3

Text Generation

alignment-handbook

Generated from Trainer

Model card Files Files and versions Metrics Training metrics Community

tinyllama_mole_dpo_ep3 / train_results.json

hushell's picture

Model save

4325427 verified 10 months ago

195 Bytes

	{
	"epoch": 1.0,
	"train_loss": 0.6456358959537526,
	"train_runtime": 17140.5232,
	"train_samples": 61135,
	"train_samples_per_second": 3.567,
	"train_steps_per_second": 0.056
	}