Minbyul
/

llama2-7b-dpo-full-wo-healthsearch_qa-ep3

Text Generation

alignment-handbook

Generated from Trainer

Inference Endpoints

text-generation-inference

Model card Files Files and versions Community

llama2-7b-dpo-full-wo-healthsearch_qa-ep3 / eval_results.json

Minbyul's picture

End of training

398deda verified 2 months ago

raw history blame contribute delete

No virus

575 Bytes

	{
	"epoch": 1.0,
	"eval_logits/chosen": 0.04453304782509804,
	"eval_logits/rejected": -0.5562469959259033,
	"eval_logps/chosen": -604.554443359375,
	"eval_logps/rejected": -867.9213256835938,
	"eval_loss": 0.6763553023338318,
	"eval_rewards/accuracies": 0.7173295617103577,
	"eval_rewards/chosen": 0.030703268945217133,
	"eval_rewards/margins": 0.03669063001871109,
	"eval_rewards/rejected": -0.005987361539155245,
	"eval_runtime": 622.4746,
	"eval_samples": 5631,
	"eval_samples_per_second": 9.046,
	"eval_steps_per_second": 0.283
	}