dunzhang

/

stella_en_1.5B_v5

like 100

What is the model loading size of this model? I am loading it 16 gb T4 gpu, it is giving cuda out of memory. 1.5B parameter should take 6GB. I am using SentenceTransformer(model_name_or_path=self.embedding_model_name, device=f'cuda:{i}')
What need to change in order to reduce model loading size?