lewtun
/

gemma-7b-dpo-full-mix1-beta-0.01

Text Generation

alignment-handbook

generated_from_trainer

Inference Endpoints

text-generation-inference

Model card Files Files and versions Community

gemma-7b-dpo-full-mix1-beta-0.01 / eval_results.json

lewtun's picture

lewtun HF staff

End of training

3319677 verified 3 months ago

raw history blame contribute delete

No virus

567 Bytes

	{
	"epoch": 0.99,
	"eval_logits/chosen": 91.75543975830078,
	"eval_logits/rejected": 85.30099487304688,
	"eval_logps/chosen": -512.5901489257812,
	"eval_logps/rejected": -565.5089721679688,
	"eval_loss": 0.5580899119377136,
	"eval_rewards/accuracies": 0.6770833134651184,
	"eval_rewards/chosen": -0.5918566584587097,
	"eval_rewards/margins": 0.5477373003959656,
	"eval_rewards/rejected": -1.1395939588546753,
	"eval_runtime": 47.3786,
	"eval_samples": 750,
	"eval_samples_per_second": 15.83,
	"eval_steps_per_second": 0.507
	}