ondevicellm
/

tinyllama_moe_dpo_ultrachat_v2_epochs3

Text Generation

alignment-handbook

Generated from Trainer

text-generation-inference

Inference Endpoints

Model card Files Files and versions Metrics Training metrics Community

tinyllama_moe_dpo_ultrachat_v2_epochs3 / train_results.json

hushell's picture

Model save

b5a6361 verified 11 months ago

194 Bytes

	{
	"epoch": 3.0,
	"train_loss": 0.5928295368507478,
	"train_runtime": 47453.5759,
	"train_samples": 61135,
	"train_samples_per_second": 3.865,
	"train_steps_per_second": 0.06
	}