How to unload AutoGPTQForCausalLM.from_quantized model to free up GPU memory

Question

I have loaded a model with the follow code:

  DEVICE = "cuda:0" if torch.cuda.is_available() else "cpu"
    print(DEVICE)

    embeddings = HuggingFaceInstructEmbeddings(
            model_name="hkunlp/instructor-xl",model_kwargs={"device":DEVICE}
    )

    model_name_or_path = "./models/Llama-2-13B-chat-GPTQ"

    model_basename = "model" 

    tokenizer = AutoTokenizer.from_pretrained(model_name_or_path, use_fast=True)

    model = AutoGPTQForCausalLM.from_quantized(
            model_name_or_path,
            revision= "gptq-4bit-128g-actorder_True", #"gptq-8bit-128g-actorder_False",  #revision="gptq-4bit-128g-actorder_True",
            model_basename=model_basename,
            use_safetensors=True,
            trust_remote_code=True,
            inject_fused_attention=False,
            device=DEVICE,
            quantize_config=None,
    )

And if I print out this model, it has the following structure:

LlamaGPTQForCausalLM(
  (model): LlamaForCausalLM(
    (model): LlamaModel(
      (embed_tokens): Embedding(32000, 5120, padding_idx=0)
      (layers): ModuleList(
        (0-39): 40 x LlamaDecoderLayer(
          (self_attn): LlamaAttention(
            (rotary_emb): LlamaRotaryEmbedding()
            (k_proj): QuantLinear()
            (o_proj): QuantLinear()
            (q_proj): QuantLinear()
            (v_proj): QuantLinear()
          )
          (mlp): LlamaMLP(
            (act_fn): SiLUActivation()
            (down_proj): QuantLinear()
            (gate_proj): QuantLinear()
            (up_proj): QuantLinear()
          )
          (input_layernorm): LlamaRMSNorm()
          (post_attention_layernorm): LlamaRMSNorm()
        )
      )
      (norm): LlamaRMSNorm()
    )
    (lm_head): Linear(in_features=5120, out_features=32000, bias=False)
  )
)

Is there a way that I can unload this model from the GPU to free up GPU memory?

I tried model.to(torch.device("cpu"))

but it did not work.

thanks