Cymela
/

hyper-3b-latent

hyper-3b-latent / hyper_config.json

Upload folder using huggingface_hub

eda3295 verified 3 months ago

139 Bytes

{
  "training_step": 82697,
  "base_model": "Qwen/Qwen2.5-3B-Instruct",
  "latent_depth_trained": 4,
  "halt_threshold_default": 0.7
}

	{
	"training_step": 82697,
	"base_model": "Qwen/Qwen2.5-3B-Instruct",
	"latent_depth_trained": 4,
	"halt_threshold_default": 0.7
	}