note : use original open llama tokenizer
model_path = 'ruwan/open-llama-sharded-1GB-7B-alpaca-vmware'
tokenizer = LlamaTokenizer.from_pretrained("openlm-research/open_llama_7b") model = LlamaForCausalLM.from_pretrained( model_path, torch_dtype=torch.float16, device_map='auto' )