ondevicellm
/

tinyllama_moe_dpo_ultrachat_v2_epochs5

Text Generation

alignment-handbook

Generated from Trainer

Inference Endpoints

text-generation-inference

Model card Files Files and versions Metrics Training metrics Community

tinyllama_moe_dpo_ultrachat_v2_epochs5 / train_results.json

hushell's picture

Model save

9ca1811 verified 5 months ago

raw history blame contribute delete

No virus

197 Bytes

	{
	"epoch": 5.0,
	"train_loss": 0.15748632995245967,
	"train_runtime": 23969.9997,
	"train_samples": 61135,
	"train_samples_per_second": 12.752,
	"train_steps_per_second": 0.199
	}